CN110284556A

CN110284556A - A kind of energy conservation based on cloud is avoided the peak hour wisdom water system

Info

Publication number: CN110284556A
Application number: CN201910612764.9A
Authority: CN
Inventors: 邓帮武; 邓卓志; 郑其元; 陈晔斌; 李威; 程建国; 张孝伟
Original assignee: Anhui Shunyu Water Affairs Co Ltd
Current assignee: Anhui Shunyu Water Affairs Co Ltd
Priority date: 2019-07-09
Filing date: 2019-07-09
Publication date: 2019-09-27
Anticipated expiration: 2039-07-09
Also published as: CN110284556B

Abstract

It supplies water the present invention relates to wisdom and manages system regions, it discloses a kind of energy conservation based on cloud to avoid the peak hour wisdom water system, it is characterized in that including pump house, water storage box, water pump assembly, water supply line component, data acquisition and transmission unit, pump house monitoring unit, wisdom water tank regulation unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit, cloud platform server and cloud platform big data analysis module.The advantage of the invention is that can be realized in non-water use peak, water storage box is carried out planning as a whole to regulate and store to optimize distribution pipe network water supply resource, it reduces in water use peak, to the i.e. hourly water demand of municipal network water supply, water supply network range of pressure fluctuations is shortened, water supply network pressure oscillation is small, the generation of pipe network model phenomenon can be reduced, guarantee that water factory's system water supply is steady, reduce the reconstruction demand of water factory's municipal network water supply and water supply line component, realizes country for the saving saving of water resource and auxiliary facility.

Description

A kind of energy conservation based on cloud is avoided the peak hour wisdom water system

Technical field

It supplies water management system regions the present invention relates to wisdom, particularly belongs to a kind of energy conservation based on cloud and avoid the peak hour wisdom confession Water system.

Background technique

Recently as the quickening of China domestic city construction process, domestic population is constantly assembled to large- and-medium size cities, So that the scale of construction of domestic large- and-medium size cities constantly increases.But large- and-medium size cities are during increase, it is necessary to abide by national phase The land policy of pass, this requires large- and-medium size cities itself must improve the effective rate of utilization in soil.Cause domestic large and medium-sized city The building for building in city develops to high-rise and small high-rise direction, and high-rise and small sized multiple story building building is above six layers.According to country The relevant regulations of urban construction: when the number of floor levels of the building for building such as each cell in city, commercial office building is greater than six layers, it is necessary to The matched secondary water-supply equipment of installation settings is to ensure to the stable water supply that builds a building.

Pipe network pressure-superposed (no negative pressure) supply equipment in 2000 or so, secondary water-supply equipment, in many urban construction It is middle to be promoted the use of a large area.But in actual use, pipe network pressure-superposed (no negative pressure) supply equipment is due to lacking water storage Ability especially when municipal Water Works scarce capacity, is sent out often so that it is higher to matched water supply infrastructure requirements Water phenomenon is robbed in life, so that the problem of water factory clashes between each other with water supply network is increasingly severe.Such as: large area uses pipe Net laminates the city of (no negative pressure) supply equipment, and in urban water peak period, Water Works transport power is not able to satisfy actual used water and needs When asking, water supply network pressure oscillation is violent, and it is serious to rob water phenomenon, so that pipe network model phenomenon happens occasionally, water factory and water supplying pipe Net reconstruction demand increases.

And for, partially using the secondary water-supply equipment with water storage box, such secondary water-supply equipment uses water non-in city When peak, water can first be stored to full water storage box, in water use peak, when the water level decreasing in water storage box is to certain altitude, this The water inlet valve of water storage box is able to maintain full-gear in class secondary water-supply equipment, continues to supply water to water storage box, until water storage Ball-cock assembly closes water inlet valve and stops intaking when case reservoir storage reaches certain liquid level.Such secondary water-supply equipment is stored by regulation Water tank realizes water supply of avoiding the peak hour of regulating and storing.But the turnover rate of water is low in such secondary water-supply equipment water storage box, it cannot be guaranteed that supplying water Water quality it is safe and reliable, not can be carried out pool regulate and store optimization distribution pipe network water supply resource, can not achieve energy conservation avoid the peak hour wisdom water supply, It is not able to satisfy the real life needs of the people.For this purpose, the present invention provides a kind of energy conservation based on cloud avoid the peak hour wisdom supply water System.

Summary of the invention

The present invention provides a kind of energy conservation based on cloud and avoids the peak hour wisdom water system, is able to solve in above-mentioned technical background The problem of mentioning, the present invention can be realized in non-water use peak, carry out planning as a whole to regulate and store to water storage box optimizing distribution pipe network water supply Resource is reduced in water use peak, to the i.e. hourly water demand of municipal network water supply, shortens municipal network water supply and water supply line group The range of pressure fluctuations of part, municipal network water supply and the fluctuation of water supply line component pressure are small, while avoiding municipal network water supply Pressure oscillation increases the instantaneous pressure of supply water of water factory, reduces the leakage loss of municipal network water supply and water supply line component, ensure that water factory The even running of municipal water supply pipeline and water supply network component reduces water factory's municipal water supply pipeline and the reconstruction of water supply network component Demand realizes country for the saving saving of water resource and auxiliary facility.

In order to achieve the above objectives, The technical solution adopted by the invention is as follows:

A kind of energy conservation based on cloud is avoided the peak hour wisdom water system, including pump house, water storage box, water pump assembly, water supplying pipe Road component, water storage box are installed in pump house, and water storage box is connect by water supply line component with water pump assembly, pump house water storage box water pump Unit has multiple, and installation settings has a water storage box in each pump house, and each water storage box is supplied by water supply line component and water pump assembly Water user connection, multiple pump houses inside water storage box are connect by water supply line component with water pump assembly, are passed through between multiple pump houses Municipal network water supply connection, which is characterized in that further include data acquisition and transmission unit, pump house monitoring unit, wisdom water tank tune Control unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit, cloud platform server With cloud platform big data analysis module, wisdom water tank regulates and controls the water quality monitoring of unit hydraulic data monitoring unit and processing unit pipe network Hydraulic data monitoring unit and pump house, water storage box, water pump assembly, water supply line component are connected with each other, and wisdom water tank regulates and controls unit The water quality monitoring of hydraulic data monitoring unit and processing unit hydraulic pipeline data monitoring unit, can by the pump house acquired in real time, The waterpower water quality information data of water storage box, water pump assembly, water supply line component, via pump house monitoring unit and data acquisition transmission Unit is transferred to Cloud Server, and instructs cloud platform big data analysis module according to Cloud Server, to more than preset standard waterpower Pump house, water storage box, water pump assembly, the water supply line component of water quality information data are adjusted, and make to be more than preset standard waterpower water Matter information data, which is adjusted to, to be complied with standard, and the cloud platform big data analysis module is embedded in cloud platform server internal, institute State data acquisition and transmission unit, pump house monitoring unit, wisdom water tank regulation unit, hydraulic data monitoring unit, water quality monitoring It is equal with the quantity of multiple pump house water storage box water pump assemblies with the quantity of processing unit, hydraulic pipeline data monitoring unit, it is described The pump house monitoring unit of each pump house can be used in sending and transmitting the regulation of the wisdom water tank in pump house unit, hydraulic data prison The monitoring data of unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit are surveyed, the pump house of each pump house monitors single Real time data is sent to cloud platform big data analysis module by the data acquisition of the pump house and transmission unit by member, and cloud platform is big The target nerve network model that data analysis module is completed based on preparatory training determines that the real time data corresponds to the target of water storage box Reservoir storage is then based on the real-time reservoir storage of corresponding water storage box, the operating state data of wisdom water tank regulation unit, supplies in real time Water number is accordingly and wisdom water tank regulation unit, the hydraulic pipeline data monitoring unit pipe network end hydraulic data that monitors determines Whether water storage is carried out to water storage box；

The pump house monitoring unit of each pump house can be used in the pump house that the hydraulic data monitoring unit of the pump house is sent It supplies water to user and is monitored with the hydraulic data of the pressure of water, flow, real-time water supply, active user water usage data；

The pump house monitoring unit of each pump house can be used in the hydraulic pipeline data monitoring unit transmission of the pump house The hydraulic data monitoring of pressure, flow, real-time water supply, real-time water-supply quantity data that pipeline supplies water to pump house water storage box；

The pump house monitoring unit of each pump house can be used in water quality monitoring to the pump house and processing unit and water quality The control of analytical equipment water disinfection equipment；

The pump house monitoring unit of each pump house can be used in the wisdom water tank regulation unit analysis water storage to the pump house Case water quality is monitored in real time and is monitored to water storage box water quality, and the water quality data information of real-time monitoring is supervised via the pump house Control unit and data acquisition transmission unit are transferred to cloud platform server and cloud platform big data analysis module, and according to cloud service Device instructs cloud platform big data analysis module to carry out disinfection processing the water body for being more than preset standard water quality, the wisdom water tank tune The data information of control unit real-time monitoring further includes the real-time reservoir storage, active user water consumption and the water that supplies water in real time of water storage box Amount；

The cloud that the cloud platform server and cloud platform big data analysis module use include public cloud, private clound and Mixed cloud, the cloud can using distributed storage, redundant storage, cold and hot backup storage by the way of memory system data, should Cloud can construct distributed computing by virtualization basic hardware resources for cloud system, effectiveness calculates, load balancing calculates, Parallel computation；

It is described should energy conservation based on cloud avoid the peak hour wisdom water system can by construct one be taken by multiple cloud platforms The unified platform for device or the storage server hardware composition of being engaged in, passes through local area network technology and connects pump house, water storage box, water pump assembly, confession Waterpipe component, computer, corollary equipment carries out data communication and transmitting instruction in pump house；

It is described should the wisdom water system of avoiding the peak hour of energy conservation based on cloud can be adopted based on public communication network infrastructure With IPsec VPN Virtual Private Network technology, the local area network based on public network is constructed, realizes corollary equipment in cloud and each pump house Data transmission；

It is described should energy conservation based on cloud avoid the peak hour wisdom water system use PLC as pump house field controller acquire show Field data and reception control instruction, and data communication is carried out by Modbus industrial bus protocol specification and cloud；

It is described should energy conservation based on cloud avoid the peak hour pump house data in wisdom water system storage by distributed data base It realizes, file process is handled by distributed file system and realized；

It is described should energy conservation based on cloud avoid the peak hour that pass through building beyond the clouds simultaneous with public communication network for wisdom water system The local area network of appearance obtains pump house data and completes cloud storage, in cloud platform server end with cloud platform big data analysis module point It analyses each pump house water data and then issues regulation and control instruction；

The cloud platform big data analysis module, for the real-time water consumption of wisdom water tank regulation unit users and each The real time data that the data acquisition of pump house and transmission unit are sent, for the real time data of each pump house, based on having trained in advance At target nerve network model determine that the real time data corresponds to the target reservoir storage of water storage box, stored when corresponding in the real time data When the real-time reservoir storage of water tank, the not up to described target reservoir storage and wisdom water tank regulation unit are not opened, sends starting and refer to It enables to the wisdom water tank and regulates and controls unit, so that wisdom water tank regulation unit starts to give water storage box water storage, and be based on the real-time number The highest reservoir storage of real-time reservoir storage, wisdom water tank regulation unit transmission water storage box in and the target water storage meter First object storage time is calculated, when reaching the first object storage time, sends the first halt instruction to the wisdom water tank Regulate and control unit, to close wisdom water tank regulation unit, when the real-time reservoir storage in the real time data is not up to the target water storage Amount and wisdom water tank regulation unit is when having turned on, it is single based on the real-time reservoir storage in the real time data, wisdom water tank regulation The specified water supply of member and the target reservoir storage calculate the second target storage time, when reaching the second target water storage Between when, send the second halt instruction to the wisdom water tank regulate and control unit so that the wisdom water tank regulation unit stop to the water storage Case water storage, when the real-time reservoir storage in the real time data has reached the target reservoir storage and has turned on wisdom water tank regulation unit It when water storage, sends third halt instruction to the wisdom water tank and regulates and controls unit, so that wisdom water tank regulation unit stops to the storage Water tank water storage, the target nerve network model be real-time water consumption based on user, the real-time water consumption sample data of user and The sample reservoir storage of corresponding water storage box is trained to obtain just as model training data, to initial target nerve network model Beginning neural network model, the target nerve network model are able to use the real-time water consumption sample data in family and corresponding water storage box Sample reservoir storage is associated, and the real-time water consumption sample data of user includes the real-time reservoir storage of sample of corresponding water storage box, sample This active user water consumption, sample cumulative water supply, the real-time water quality data of sample and the real-time water supply of sample.

Preferably, the cloud platform big data analysis module, additionally it is possible in the target mind completed based on preparatory training Before determining the target reservoir storage that the real time data corresponds to water storage box through network model, based on discrete cosine transformation matrix to the reality When data carry out discrete cosine transform obtain the real time data after de-redundancy, the mean value of the real time data after calculating the de-redundancy And standard deviation, the real time data after the de-redundancy is normalized based on mean value described in this and the standard deviation Real time data after to normalization, it is real-time after the normalization is determined based on the preparatory target nerve network model for training completion Data correspond to the target reservoir storage of water storage box.

Preferably, the cloud platform big data analysis module specifically can be used in that water storage box will be corresponded in real time data The sample point data that real-time reservoir storage, active user water consumption, accumulation water supply, real-time water quality data and real-time water supply include It is arranged in the first matrix, the product for calculating first matrix and discrete cosine transformation matrix obtains the second matrix, by described the Each element in two matrixes is as the real time data for corresponding to pump house after de-redundancy.

Preferably, the described cloud platform big data analysis module, specifically can also be used to calculate table in second matrix The first mean value and the first standard deviation of the first element of reservoir storage calculate this first yuan for each first element when levies in kind First difference of corresponding first mean value of element, first difference of calculating and the first quotient of first standard deviation Value；

And calculate the second mean value and the second mark for the second element that active user water consumption is characterized in second matrix It is quasi- poor, for each second element, the second difference of corresponding second mean value of the second element is calculated, described in calculating Second quotient of the second difference and second standard deviation；

And the third mean value and third standard deviation of the third element of characterization accumulation water supply in second matrix are calculated, For each third element, the third difference of the corresponding third mean value of the third element is calculated, the third is calculated The third quotient of difference and the third standard deviation；

And calculate the 4th mean value and the 4th standard for the fourth element that real-time water quality data is characterized in second matrix Difference calculates the 4th difference of corresponding the 4th mean value of the fourth element for each fourth element, calculates described the 4th quotient of four differences and the 4th standard deviation；

And the 5th mean value and the 5th standard deviation that the The Fifth Element of real-time water supply is characterized in second matrix are calculated, For each The Fifth Element, the 5th difference of corresponding the 5th mean value of the The Fifth Element is calculated, calculates the described 5th 5th quotient of difference and the 5th standard deviation；

First quotient, second quotient, the third quotient, the 4th quotient and the 5th quotient are made For the real time data after the normalization.

Preferably, the training process of the target nerve network model, specifically:

Obtain the sample reservoir storage of the real-time water consumption sample data of user and corresponding water storage box that training uses, the user Real-time water consumption sample data includes the real-time reservoir storage of sample of corresponding water storage box, sample active user water consumption, sample cumulative The real-time water quality data of water supply, sample and the real-time water supply of sample；

The sample reservoir storage of the real-time water consumption sample data of the user and corresponding water storage box is inputted into initial neural network In model, the initial neural network model includes input layer, hidden layer, output layer and establishes in input layer and output layer Between equation group relationship, the equation group relationship of the foundation between input layer and output layer is excitation function；

It is respectively saved according to the real-time water consumption sample data of the user and the sample reservoir storage of corresponding water storage box, the input layer The output node layer of weight coefficient, each node of the hidden layer and the output layer between point and each node of the hidden layer it Between weight coefficient and the equation group relationship of the foundation between the input layer and the output layer；

The equation group is solved, the weight coefficient between each node of the input layer and each node of the hidden layer is obtained The value of weight coefficient between the output node layer of value, each node of the hidden layer and the output layer；

The value of weight coefficient between each node of the input layer and each node of the hidden layer, the hidden layer are respectively saved The value and the real-time water consumption sample data of the user of weight coefficient between point and the output node layer of the output layer substitute into In the equation group, the initial reservoir storage that the real-time water consumption sample data of the user corresponds to water storage box is solved；

Calculate the difference between the initial reservoir storage of the corresponding water storage box and the sample reservoir storage of the corresponding water storage box Value；

When the difference value is greater than default discrepancy threshold, first weight coefficient and institute are adjusted based on the difference value The second weight coefficient is stated, returns to execution the sample reservoir storage of the real-time water consumption sample data of the user and corresponding water storage box is defeated Enter the step in initial neural network model, first weight coefficient is that first input layer is hidden with described first The first weight coefficient between node layer, second weight coefficient are second input layer and first hidden layer The second weight coefficient between node；

When the difference value is not more than default discrepancy threshold, training is completed, is obtained comprising the real-time water consumption sample of user The target nerve network model of data and the corresponding relationship of the sample reservoir storage of the corresponding water storage box.

Preferably, the input layer includes the first input layer, the second input layer and third input layer, institute Stating hidden layer includes that the first hiding node layer, the second hiding node layer and third hide node layer, and the output layer includes output Node layer, it is described according to the real-time water consumption sample data of the user and the sample reservoir storage of corresponding water storage box, the input layer The output layer section of weight coefficient, each node of the hidden layer and the output layer between each node and each node of the hidden layer The step of weight coefficient between point and the equation group relationship of the foundation between the input layer and the output layer, packet It includes:

According to the real-time water consumption sample data of the user, first input layer and the described first hiding node layer Between the first weight coefficient, the second weight coefficient between second input layer and the first hiding node layer and The first constant between the third input layer and the first hiding node layer establishes the first linear equation；

According to the real-time water consumption sample data of the user, first input layer and the described second hiding node layer Between third weight coefficient, the 4th weight coefficient between second input layer and the second hiding node layer and The second constant between the third input layer and the second hiding node layer establishes the second linear equation；

Node layer is hidden according to the real-time water consumption sample data of the user, first input layer and the third Between the 5th weight coefficient, second input layer and the third hide the 6th weight coefficient between node layer and The third constant that the third input layer and the third are hidden between node layer establishes third linear equation；

It is excitation function, the use according to equation group relationship of the foundation between the input layer and the output layer The real-time water consumption sample data in family corresponds to the sample reservoir storage of water storage box, the functional value of first linear equation, described second The functional value of the functional value of linear equation and the third linear equation, the first hiding node layer and the output node layer Between the 7th weight coefficient, the second hiding node layer and it is described output node layer between the 8th weight coefficient and institute It states the 9th weight coefficient that third is hidden between node layer and the output node layer and establishes the 4th functional equation；

First linear equation described in simultaneous, second linear equation, the third linear equation and the 4th function Equation obtains the equation group between the input layer and the output layer；

It is described to solve the equation group, obtain the weight system between each node of the input layer and each node of the hidden layer The step of value of weight coefficient between the output node layer of several value, each node of the hidden layer and the output layer, comprising:

The equation group is solved, the value of first weight, the value of second weight, the third weight are obtained Value, the value of the 4th weight, the value of the 5th weight, the value of the 6th weight, the value of the 7th weight, described The value of the value of eight weights and the 9th weight；

It is described by the value of the weight coefficient between each node of the input layer and each node of the hidden layer, the hidden layer The value and the real-time water consumption sample data of the user of weight coefficient between each node and the output node layer of the output layer It substitutes into the equation group, solves the step that the real-time water consumption sample data of the user corresponds to the initial reservoir storage of water storage box Suddenly, comprising:

By the value of first weight, the value of second weight, the value of the third weight, the 4th weight Value, the value of the 5th weight, the value of the 6th weight, the value of the 7th weight, the value of the 8th weight and described The value of 9th weight and the real-time water consumption sample data of the user substitute into the equation group, solve the user and use in real time Water sample data corresponds to the initial reservoir storage of water storage box.

Preferably, the functional value of the excitation function are as follows: the sum of drive factor and target max, the target max For the maximum value among the real-time water consumption sample data of the user and target product, the target product is excitation system Several products with the real-time water consumption sample data of the user.

Preferably, also installation settings has sterilizer and water storage box self cleaning arrangement, institute in the water storage box of each pump house State cloud platform big data analysis module, additionally it is possible to for determining the storage of the pump house according to the real-time water quality data in the real time data Whether the water quality in water tank reaches preset standard, when the duration that water quality is not up to preset standard reaches the first preset duration When, the sterilizer in sterilizer open command to the pump house water storage box is sent, so that the sterilizer is opened, when water quality is not up to When the duration of preset standard reaches the second preset duration, water storage box self cleaning arrangement open command is sent to the pump house water storage Water storage box self cleaning arrangement in case, so that the water storage box self cleaning arrangement is opened, first preset duration is less than described Second preset duration.

Preferably, the real-time water quality data is real-time water quality, the real-time water quality data according in the real time data The step of whether water quality in the water storage box of the pump house reaches preset standard determined, comprising:

Judge whether the real-time water quality is less than predetermined quality, if so, determining that the water quality in the water storage box of the pump house reaches To preset standard, if not, determining that the water storage water quality of the pump house is not up to preset standard.

Compared with the prior art, beneficial effects of the present invention are as follows:

It is single by being monitored to pump house, water storage box, water pump assembly, water supply line component, data acquisition and transmission unit, pump house Member, wisdom water tank regulation unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit, The global optimization R & D design of cloud platform server and cloud platform big data analysis module is manufactured that a kind of based on cloud Energy conservation is avoided the peak hour wisdom water system.The pump house monitoring unit of each pump house in the water system, can pass through real time data The data acquisition of the pump house and transmission unit are sent to cloud platform big data analysis module, and each wisdom water tank regulation unit is sent The maximum water-storage of corresponding water storage box to cloud platform big data analysis module, cloud platform big data analysis module is based on preparatory The target nerve network model that training is completed, determines that the real time data corresponds to the target reservoir storage of water storage box；Then based on real When reservoir storage, wisdom water tank regulation unit determine corresponding water storage box maximum water requirement, target reservoir storage and wisdom Whether water tank regulation unit has started to supply water to water storage box, reaches to control wisdom water tank regulation unit in real-time reservoir storage When before target reservoir storage to water storage box supply water, realize in non-water use peak, optimization distribution pipe network water supply resource to water storage box into Row is planned as a whole to regulate and store, and reduces the hourly water demand that water supply network supplies water in water use peak, to realize the purpose for water supply of avoiding the peak hour.Into And realize the optimum use to water supply network facility, it realizes to country for the saving saving of water resource and auxiliary facility.But It implements any of the products of the present invention or method does not necessarily require achieving all the advantages described above at the same time.Specific beneficial effect It is as follows:

1) the pump house monitoring unit of each pump house sends real time data by the data acquisition of the pump house and transmission unit To cloud platform big data analysis module, target nerve network mould of the cloud platform big data analysis module based on preparatory training completion Type determines that the real time data corresponds to the target reservoir storage of water storage box；Unit is then being regulated and controled based on real-time reservoir storage, wisdom water tank Whether maximum water requirement, target reservoir storage and the wisdom water tank regulation unit of determining corresponding water storage box have started to storage Cistern water supply is supplied when real-time reservoir storage reaches before target reservoir storage to water storage box to control wisdom water tank regulation unit Water carries out planning as a whole to regulate and store, reduce in water use peak to meet the optimization distribution pipe network water supply resource in non-water use peak to water tank When municipal network water supply supply water hourly water demand, shorten the range of pressure fluctuations of municipal network water supply and water supply line component, Municipal network water supply and the fluctuation of water supply line component pressure are small, while the pressure oscillation of municipal network water supply being avoided to increase water factory's wink When pressure of supply water, reduce the leakage loss of pipe network, ensure that Water Works system even running, reduce water factory's municipal network water supply and The reconstruction demand of water supply line component realizes country for the saving saving of water resource and auxiliary facility；

2) wisdom water tank regulation unit can connect cloud platform server, receive the finger that cloud platform server plans as a whole water storage It enables, water storage of avoiding the peak hour is carried out to pump house water storage box；

3) de-redundancy is carried out to real time data based on discrete cosine transformation matrix, improves subsequent prediction target reservoir storage Accuracy can be improved the calculating of subsequent prediction target reservoir storage by the way that the real time data after de-redundancy is normalized Convenience；

4) by being trained to obtain initial neural network model to initial target nerve network model, then by mating The training method initial neural network model is trained, can obtain so that the real-time water consumption sample data of user It, can be pre- by the target nerve network model with the associated target nerve network model of the sample reservoir storage of corresponding water storage box Target reservoir storage is surveyed, is pumped with corresponding water storage box to corresponding water pump assembly in order to subsequent based on the control of target reservoir storage Or stop pumping, achieve the purpose that reduce energy waste, be not necessarily to manual control water pump assembly, reaches and automatically control the water system Energy conservation is regulated and stored the purpose of operation, has been saved cost of labor, has been improved work efficiency；

5) use equation group relationship of the heretofore described foundation between input layer and output layer for excitation function, it is described Excitation function can exclude in the real-time water consumption sample data of user, greater than the real-time water consumption sample of the user of negative drive factor Data, usually the water system in actual use, it is believed that be greater than in the real-time water consumption sample data of user and negative swash The real-time water consumption sample data of user for encouraging coefficient is the excessive data of noise, by the real-time water consumption of the excessive user of these noises Sample data excludes, and can reduce or eliminate noise to the stability of the target nerve network model in the water system, cause Adverse effect, while the prediction error of target nerve network model in the water system can be reduced.

6) also installation settings has sterilizer and water storage box self cleaning arrangement, such mode in the water storage box of each pump house When can correspond to the water quality of water storage box in pump house and being not up to preset standard, control or instruction sterilizer and water storage box automatically cleaning are set It is standby to carry out disinfection and cleaning treatment to the water body in water storage box, to ensure the safety compliance of user's water supply quality.

Detailed description of the invention

Fig. 1 is the cloud platform server of water system medium cloud plateform system of the present invention, the cloud for being embedded in cloud platform server Platform big data analysis module and water storage box, data acquisition and the transmission unit, pump house monitoring unit, intelligence for being set to each pump house Connect between intelligent water tank regulation unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit The model configuration schematic diagram connect；

Fig. 2 is that the overall region for each pump house being arranged in water system of the present invention and the virtual network of cloud platform system connect Connect structural schematic diagram；

Fig. 3 is group in the specific hardware support kit facility and mating cloud platform system of a pump house in water system of the present invention Structural schematic diagram interconnected between part；

Fig. 4 is that water system medium cloud plateform system target nerve network model internal trainer process association structure of the present invention shows It is intended to.

Specific embodiment

Following will be combined with the drawings in the embodiments of the present invention and Detailed description of the invention, technical solution in the embodiment of the present invention into Clear, the complete description of row.Obviously, described embodiment is only a part of the embodiments of the present invention, rather than whole Embodiment.Based on the embodiments of the present invention, those of ordinary skill in the art are obtained under that premise of not paying creative labor The every other embodiment obtained, shall fall within the protection scope of the present invention.

It should be noted that term " includes " and " having " and their times in the embodiment of the present invention and Detailed description of the invention What is deformed, it is intended that is covered and non-exclusive is included.Such as comprising a series of steps or units process, method, system, production Product or equipment are not limited to listed step or unit, but preferably further include the steps that not listing or unit, or It preferably further include other step or units intrinsic for these process, methods, product or equipment.

It avoids the peak hour wisdom water system the invention discloses a kind of energy conservation based on cloud, it can be by cloud platform service The cloud platform of device and matching network infrastructure composition, plans as a whole the storage in the water storage box and water storage box in control area in each pump house Water realizes that optimization distribution pipe network water supply resource carries out planning as a whole to regulate and store to water tank in non-water use peak, reduces in water use peak When municipal network water supply supply water hourly water demand, shorten the range of pressure fluctuations of municipal network water supply and water supply line component, Municipal network water supply and the fluctuation of water supply line component pressure are small, while the pressure oscillation of municipal network water supply being avoided to increase water factory's wink When pressure of supply water, reduce the leakage loss of pipe network, ensure that Water Works system even running, reduce water factory's municipal network water supply and The reconstruction demand of water supply line component realizes country for the saving saving of water resource and auxiliary facility.It below will be of the invention Specific embodiment, which is described with reference to the drawings, to be described in detail, specific as follows, referring to attached drawing:

Fig. 1 is the cloud platform server of water system medium cloud plateform system of the present invention, the cloud for being embedded in cloud platform server Platform big data analysis module and water storage box, data acquisition and the transmission unit, pump house monitoring unit, intelligence for being set to each pump house Connect between intelligent water tank regulation unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit The model configuration schematic diagram connect；It is specifically included in Fig. 1: the cloud platform server (1) of water system medium cloud plateform system of the present invention, It is embedded in the cloud platform big data analysis module (2) of cloud platform server (1), each pump house of cloud platform server (1) control (3), the internal water storage box (4) of each pump house (3), data acquisition and transmission unit (5), pump house monitoring unit (6), intelligence are set to Intelligent water tank regulation unit (7), hydraulic data monitoring unit (8), water quality monitoring and processing unit (9), hydraulic pipeline data monitoring Unit (10).

Illustrate the overall region and cloud of each pump house being arranged in water system of the present invention in order to vivider expression The virtual network connection of plateform system has used Fig. 2 expression (referring specifically to Fig. 2)；Illustrate this hair in order to vivider expression Being connected with each other between component in the specific hardware support kit facility of a pump house in bright water system and mating cloud platform system makes It is indicated with Fig. 3 (referring specifically to Fig. 3)；By Fig. 2 and Fig. 3 can be vivider indicate to illustrate it is provided by the invention based on cloud Energy conservation avoid the peak hour wisdom water system structure and cloud platform server (1) control each pump house (3) in it is specific mating Connection relationship between equipment.

The present invention is in actual use operational process, in order to each region for using water system of the present invention in city Wisdom water tank regulation unit (7) in each pump house (3) being arranged carries out intelligent control, and mating be provided with pumps in the region The cloud platform server (1) in room (3), to manage the specific corollary equipment in each pump house (3) realize energy conservation avoid the peak hour wisdom supply The overall control center for being mounted on city entirety water system can be set in water running, the cloud platform server (1).Cloud platform clothes Business device (1) is embedded with cloud platform big data analysis module (2), and cloud platform big data analysis module (2) can pass through wireless local area Net is communicatively coupled with the specific corollary equipment being arranged in pump house.Wisdom water tank regulation unit (7) hair of each pump house (3) It is given to manage the maximum water-storage of corresponding water storage box (4) to cloud platform big data analysis module (2)；

The water storage box (4) being arranged in each pump house (3), and wisdom water tank corresponding with water storage box (4) regulate and control unit (7), wisdom water tank regulation unit (7) can control municipal network water supply and water supply line component, be stored to water storage box (4) Water or without water storage, water storage box (4) can save the water body to supply water to user, and water storage box (4) can pass through multiple water supplying pipes Road component supplies water to different users.Meanwhile water storage box (4) is equipped with security protection door above, is realized by security protection door to water storage The security protection of case (4)；

Wisdom water tank regulation unit (7) of each pump house (3) can also obtain corresponding water storage box (4) in the pump house (3) Real-time reservoir storage, and acquired real-time reservoir storage is sent to the pump house monitoring unit (6) of the pump house (3)；Foundation simultaneously The finger from cloud platform big data analysis module (2) of wisdom water tank regulation unit (7) is sent to by pump house monitoring unit (6) It enables starting or stops carrying out water storage to the water storage box (4)；

The hydraulic data monitoring unit (8) of each pump house (3) can obtain the waterpower number that the pump house (3) supplies water to user According to the hydraulic data including pressure of supply water and water supply flow, water supply flow is active user water consumption, and is sent the data to The pump house monitoring unit (6)；

The water quality monitoring of each pump house (3) and processing unit (9) can obtain corresponding water storage box (4) in the pump house (3) Real-time water quality data, and acquired real-time water quality data is sent to the pump house monitoring unit (6) of the pump house (3)；

The hydraulic pipeline data monitoring unit (10) of each pump house (3) can obtain municipal network water supply into the pump house (3) hydraulic data of end includes pressure of supply water and water supply flow, and water supply flow is that municipal network water supply is right into the pump house (3) The real-time water supply for the water storage box (4) answered, and send the data to the pump house monitoring unit (6)；

The pump house monitoring unit (6) of each pump house (3), can receive the pump house (3) wisdom water tank regulation unit (7), The pump that hydraulic data monitoring unit (8), water quality monitoring and processing unit (9) and hydraulic pipeline data monitoring unit (10) are sent Real-time pressure of supply water data that room (3) supplies water to user, water supply flow data (as active user water usage data), Real-time Water The hydraulic data that prime number evidence and municipal network water supply enter the pump house end includes pressure of supply water, water supply flow (as municipal administration confession The real-time water supply of pipe network corresponding water storage box (4) into the pump house (3)) data, and these real time datas are passed through into number Cloud platform big data analysis module (2) is sent to according to acquisition and transmission unit (5)；

Cloud platform big data analysis module (2) can receive the regulation of the wisdom water tank in each pump house (3) unit (7), water The pump house that force data monitoring unit (8), water quality monitoring and processing unit (9) and hydraulic pipeline data monitoring unit (10) are sent Real-time pressure of supply water data, the water supply flow data (as active user water usage data), Real-time Water prime number to supply water to user According to and municipal network water supply enter the pump house end hydraulic data include pressure of supply water, water supply flow (as municipal water supply pipe The real-time water supply of net corresponding water storage box (4) into the pump house (3)) data；

Since the active user water consumption of each pump house (3) has certain periodicity and regularity, that is, each pump The reservoir storage of the corresponding water storage box (4) in room (3), it is only necessary to meet the water demand of subsequent period user；It does not need Reach the reservoir storage when water storage box (4) stores full water, when the reservoir storage of the water storage box (4) meets user's water demand, Yun Ping Platform big data analysis module (2) can control corresponding wisdom water tank regulation unit (7) in this region and stop to the correspondence Water storage box (4) carry out water storage.Therefore, cloud platform big data analysis module (2) is in the real time data for receiving each pump house (3) Afterwards, it needs to predict the target reservoir storage that the real time data corresponds to water storage box (4) according to the real time data of each pump house (3).

In order to predict that the real time data corresponds to the target reservoir storage of water storage box (4), trained in advance in water system of the present invention Target nerve network model.Target nerve network model in water system of the present invention are as follows: water is used based on user in real time Training data of the sample reservoir storage of amount, the real-time water consumption sample data of user and corresponding water storage box (4) as the model, to first The target nerve network model of beginning is trained to obtain initial neural network model, and target nerve network model is able to use family reality Hourly water consumption sample data is associated with the sample reservoir storage of corresponding water storage box (4), and the real-time water consumption sample data of user includes The real-time reservoir storage of sample, sample active user water consumption, sample cumulative water supply, the real-time water quality of sample of corresponding water storage box (4) Data and the real-time water supply of sample.

The training process of target nerve network model is specifically introduced, as follows:

The training process of target nerve network model can are as follows:

The sample data for the real-time water consumption of user that training uses is obtained, sample data includes the sample of corresponding water storage box (4) Reservoir storage, sample active user water consumption, sample cumulative water supply, the real-time water quality data of sample and sample supply water in real time in real time for this Amount；

Sample data and real-time active user water consumption are inputted in initial neural network model, initial neural network model Equation group relationship including input layer, hidden layer, output layer and foundation between input layer and output layer is established in input layer Equation group relationship between output layer is excitation function；

According to sample reservoir storage, each node of input layer and each node of hidden layer of sample data and corresponding water storage box (4) it Between weight coefficient, the weight coefficient between each node of hidden layer and the output node layer of output layer and establish input layer and output The excitation function of equation group relationship between layer；

Equation group is solved, value, the hidden layer of the weight coefficient between each node of input layer and each node of hidden layer are obtained The value of weight coefficient between each node and the output node layer of output layer；

By the value of the weight coefficient between each node of input layer and each node of hidden layer, each node of hidden layer and output layer The value and the real-time water consumption sample data of user for exporting the weight coefficient between node layer substitute into equation group, solve real-time use Family water consumption corresponds to the initial reservoir storage of water storage box (4)；

Calculate the difference value between initial reservoir storage and target reservoir storage；

When difference value is greater than default discrepancy threshold, the first weight and the second weight are adjusted based on difference value, returns and executes The sample reservoir storage of the real-time water consumption sample data of user and corresponding water storage box (4) is inputted into the step in initial neural network model Suddenly；First weight system of first weight coefficient between first input layer and the first hiding node layer Number, second weight system of second weight coefficient between second input layer and the first hiding node layer Number；

When difference value is not more than default difference value threshold value, training is completed, is obtained comprising the real-time water consumption sample number of user According to the target nerve network model of the corresponding relationship of the sample reservoir storage of corresponding water storage box (4).

When establishing target nerve network model, need to obtain the sample number of the real-time water consumption of user used in training set According to and corresponding water storage box (4) sample reservoir storage, the real-time water consumption sample data of user includes that the sample of corresponding water storage box (4) is real When reservoir storage, sample active user water consumption, sample cumulative water supply, the real-time water quality data of sample and the real-time water supply of sample； The sample data of the real-time water consumption of user used in training set and the sample reservoir storage of corresponding water storage box (4), for for difference Pump house (3) in different time the collected data of institute.

A kind of energy conservation based on cloud of the present invention is avoided the peak hour the initial neural network model in wisdom water system, additionally it is possible to Be understood to: the associated electronic device of specific corollary equipment needs to construct an initial neural network model in the present invention, to first Beginning neural network model is trained, and obtains target nerve network model.In a kind of wherein mode that can be implemented, utilize Caffe tools build one includes the equation group of input layer, hidden layer, output layer and foundation between input layer and output layer The initial neural network model of excitation function relationship；

Present system, can be by reality after obtaining the real-time water consumption sample data of user and corresponding sample reservoir storage When user's water usage data and the real-time reservoir storage of corresponding water storage box (4) input in initial neural network model, the initial nerve Network model includes the equation group excitation function of input layer, hidden layer, output layer and foundation between input layer and output layer Relationship；

Then it according to active user water consumption and the real-time reservoir storage of corresponding water storage box (4), each node of input layer and hides It weight coefficient, each node of hidden layer between each node of layer and the weight coefficient between the output node layer of the output layer and builds Equation group excitation function relationship between vertical input layer and output layer, establishes equation group, at this point, each node of input layer and hidden layer Weight coefficient between each node of weight coefficient, hidden layer between each node and the output node layer of the output layer is unknown Number；

After establishing equation group, solve system of equation obtains the power between each node of input layer and each node of hidden layer Value, the value for exporting the weight coefficient between node layer of each node of hidden layer and the output layer of weight coefficient；Input layer is each Between the value of weight coefficient between node and each node of hidden layer, each node of hidden layer and the output node layer of the output layer Weight coefficient value and the real-time water consumption sample data of user substitute into equation group, solve that active user water consumption is corresponding to be stored The initial reservoir storage of water tank (4)；

Between the initial reservoir storage and sample reservoir storage for calculating corresponding water storage box (4) by objective function predetermined Difference value illustrates that initial neural network model at this time also fails to be suitable for big portion when difference value is greater than default discrepancy threshold The real-time water consumption sample data of user divided, at this time, it may be necessary to which passing through back propagation according to difference value adjusts the first weight coefficient With the second weight coefficient, execution is returned to the sample reservoir storage of the real-time water consumption sample data of user and corresponding water storage box (4) is defeated Enter the step in initial neural network model；

In the training process, present system can loop through each region that water system of the present invention is used in city The real-time water consumption sample data of all users, and constantly adjust the first weight coefficient and the second weight coefficient, when difference value not When greater than default discrepancy threshold, illustrate that initial neural network model at this time can be suitable for the real-time water consumption of most user Sample data obtains accurately as a result, at this point, completion training, obtains storing comprising the real-time water consumption sample data of user with corresponding The target nerve network model of the corresponding relationship of water tank (4) sample reservoir storage.

A kind of energy conservation based on cloud of the present invention is avoided the peak hour the initial neural network model in wisdom water system, can be by It is specific again to understand are as follows: input layer includes the first input layer, the second input layer and third input layer, hidden layer packet It includes the first hiding node layer, the second hiding node layer and third and hides node layer, output layer includes output node layer, according to user Between real-time water consumption sample data and the sample reservoir storage of corresponding water storage box (4), each node of input layer and each node of hidden layer Weight coefficient, each node of hidden layer and output layer output node layer between weight coefficient and establish in input layer and output Layer between equation group excitation function the step of, can include:

According to the first power between the real-time water consumption sample data of user, the first input layer and the first hiding node layer The second weight coefficient and third input layer and first between weight coefficient, the second input layer and the first hiding node layer The first constant hidden between node layer establishes the first linear equation；

According to the third power between the real-time water consumption sample data of user, the first input layer and the second hiding node layer The 4th weight coefficient and third input layer and second between weight coefficient, the second input layer and the second hiding node layer The second constant hidden between node layer establishes the second linear equation；

The 5th power between node layer is hidden according to the real-time water consumption sample data of user, the first input layer and third The 6th weight coefficient and third input layer and third between the hiding node layer of weight coefficient, the second input layer and third The third constant hidden between node layer establishes third linear equation；

According to the equation group excitation function relationship between input layer and output layer of foundation, the real-time water consumption sample number of user According to corresponding sample reservoir storage, the functional value of the first linear equation, the functional value of the second linear equation and third linear equation The 7th weight coefficient, the second hiding node layer and output layer section between functional value, the first hiding node layer and output node layer The 9th weight coefficient that the 8th weight coefficient and third between point are hidden node layer and exported between node layer establishes the 4th Functional equation；

The first linear equation of simultaneous, the second linear equation, third linear equation and the 4th functional equation, obtain input layer with Equation group between output layer；

The equation group is solved, the value of the weight coefficient between each node of input layer and each node of hidden layer is obtained, hides The step of value of weight coefficient between each node of layer and the output node layer of output layer, comprising:

The equation group is solved, the value of the first weight, the value of the second weight, the value of third weight, the 4th weight are obtained Value, the value of the 5th weight, the value of the 6th weight, value, the value of the 8th weight and the value of the 9th weight of the 7th weight；

By the value of the weight coefficient between each node of input layer and each node of hidden layer, each node of hidden layer and output layer The value and the real-time water consumption sample data of user for exporting the weight coefficient between node layer substitute into equation group, solve user's reality Hourly water consumption sample data corresponds to the step of initial reservoir storage of water storage box (4), comprising:

By the value of the first weight, the value of the second weight, the value of third weight, the value of the 4th weight, the value of the 5th weight, The value of six weights, the value of the 7th weight, the value of the 8th weight and the value of the 9th weight and user's real-time water consumption sample data generation Enter in equation group, solves the initial reservoir storage that the real-time water consumption sample data of user corresponds to water storage box (4).

The exemplary embodiment (referring to fig. 4) of specific embodiment, Fig. 4 are water system medium cloud plateform system target of the present invention Neural network model internal trainer process association structure schematic diagram.From fig. 4, it can be seen that three, the left side of target nerve network model section Point X1 is the first input layer, X2 is that the second input layer and 1 is third input layer, and right side node y is output layer Node, h1 is the first hiding node layer, h2 is the second hiding node layer, h3 is that third hides node layer, σ indicates excitation function, Each node is also referred to as neuron, is connected between neuron and forms target nerve network, deposits between input layer and hidden layer In corresponding weight coefficient and constant, there are corresponding weight coefficient, the target nerve network moulds between hidden layer and output layer Type is by excitation function (establishing the equation group excitation function relationship between input layer and output layer), weight coefficient, constant and mind It is determined through the connection type between member, the connection type of target nerve network model is full connection, i.e., between adjacent two layers It is connected between any two node, the advantage of doing so is that the potentially relevant pass between different types of data can be excavated System.

Relationship between the input and output of target nerve network model in Fig. 4 can be defined by following equation:

a₁=w_1-11x₁+w_1-21x₂+b_1-1

a₂=w_1-12x₁+w_1-22x₂+b_1-2

a₃=w_1-13x₁+w_1-23x₂+b_1-3

Y=σ (w_2-1σ(a₁)+w_2-2σ(a₂)+w_2-3σ(a₃))

Wherein, a₁For the functional value of the first linear equation, a₂For the functional value of the second linear equation, a₃For third linear side The functional value of journey, w_1-11For the first weight coefficient between the first input layer and the first hiding node layer, w_1-21It is defeated for second Enter the second weight coefficient between node layer and the first hiding node layer, w_1-12For the first input layer and the second hidden layer section Third weight coefficient between point, w_1-22For the 4th weight coefficient between the second input layer and the second hiding node layer, w_1-13The 5th weight coefficient between node layer, w are hidden for the first input layer and third_1-23For the second input layer with Third hides the 6th weight coefficient between node layer, w_2-1The 7th power between the first hiding node layer and output node layer Weight coefficient, w_2-2For the 8th weight coefficient between the second hiding node layer and output node layer, w_2-3Node layer is hidden for third With the 9th weight coefficient between output node layer, b_1-1For first between third input layer and the first hiding node layer Constant, b_1-2For the second constant between third input layer and the second hiding node layer, b_1-3For third input layer with Third hides the third constant between node layer, and y is the corresponding sample reservoir storage of the real-time water consumption sample data of user, σ (x_i) For about x_iExcitation function, x_iFor the real-time water consumption sample data of i-th of user, x₁For the 1st real-time water consumption sample number of user According to x₂For the 2nd real-time water consumption sample data of user, a₁=w_1-11x₁+w_1-21x₂+b_1-1For the first linear equation, a₂=w_1- ₁₂x₁+w_1-22x₂+b_1-2For the second linear equation a₃=w_1-13x₁+w_1-23x₂+b_1-3For third linear equation, y=σ (w_2-1σ(a₁)+ w_2-2σ(a₂)+w_2-3σ(a₃)) it is the 4th functional equation.

The functional value of excitation function can in above-mentioned exemplary embodiment are as follows: the sum of drive factor and target max, it is described Target max is the maximum value among the real-time water consumption sample data of user and target product, and target product is drive factor With the product of the real-time water consumption sample data of user.Drive factor in above formula is -0.1, the real-time water consumption sample data of user For the real-time water consumption sample data of i-th of user.User can be excluded using the excitation function in the embodiment of the present invention to use in real time It is greater than the real-time water consumption sample data of user of negative drive factor in water sample data, it is usually practical in the water system In use process, it is believed that be greater than the real-time water consumption sample number of user of negative drive factor in the real-time water consumption sample data of user According to the data excessive for noise, the real-time water consumption sample data of the excessive user of these noises is excluded, can be reduced or eliminated Noise to the stability of the target nerve network model in the water system, caused by adversely affect, while the confession can be reduced The prediction error of target nerve network model in water system.

As following table gives using standard ReLU (Rectified Linear Unit, line rectification function) excitation function Performance with the excitation function in the embodiment of the present invention in prediction target reservoir storage compares, and the numerical value in table indicates mean error Value, smaller expression performance is higher, is predicted by the visible excitation function using the embodiment of the present invention of table when predicting target reservoir storage Error is smaller, and accuracy rate is higher.

Excitation function	Data set 1	Data set 2	Data set 3
				The embodiment of the present invention	5.2%	2.3%	10.1%
Standard ReLU	10.5%	5.5%	13.7%

In conclusion being trained by above-mentioned training method to initial neural network model, can obtain so that user Real-time water consumption sample data and the associated target nerve network model of the sample reservoir storage of corresponding water storage box, pass through the target Neural network model can predict the target reservoir storage of corresponding water storage box to get the watering law feature of user is arrived, in order to The subsequent target reservoir storage based on corresponding water storage box controls the water pump assembly in the pump house and is pumped or stop pumping, and reaches and subtracts The purpose of few energy waste, meanwhile, it is not necessarily to manual control water pump assembly, has reached remote auto control (as cloud control) Purpose, saved cost of labor, improved work efficiency.

Cloud platform big data analysis module (2) is based on the real time data of each pump house (3) in water system of the present invention The target nerve network model that training is completed in advance determines that the real time data corresponds to the target reservoir storage of water storage box (4), when the reality When data in correspond to water storage box (4) real-time reservoir storage miss the mark reservoir storage and the pump house (3) the regulation of wisdom water tank it is single When first (7) do not open to water storage box (4) progress water storage, illustrate that the reservoir storage in water storage box (4) is not corresponding with pump house (3) enough User use, need wisdom water tank regulation unit (7) accuse municipal network water supply and water supply line component to water storage box (4) into Row water storage；At this point, the wisdom water tank that cloud platform big data analysis module (2) sends enabled instruction to the pump house (3) regulates and controls unit (7), so that wisdom water tank regulation unit (7) of the pump house (3) accuses that municipal network water supply and water supply line component start to storage Water tank (4) carries out water storage；

In order to save the energy, when the reservoir storage in water storage box (4) reaches target reservoir storage, that is, stop to water storage box (4) Water storage.At this point, cloud platform big data analysis module (2) can be based on the real-time water storage for corresponding to water storage box (4) in the real time data The maximum water requirement of the water storage box (4) that wisdom water tank regulation unit (7) of amount, the pump house (3) stores and target reservoir storage calculate First object storage time, when reaching first object storage time, cloud platform big data analysis module (2) sends first and stops Instruct to the pump house (3) wisdom water tank regulate and control unit (7) so that the pump house (3) wisdom water tank regulation unit (7) stop to Corresponding water storage box (4) carries out water storage.

Wherein, it is based on corresponding to the real-time reservoir storage of water storage box (4), the wisdom water tank tune of the pump house (3) in the real time data The maximum water-storage and target reservoir storage for controlling the water storage box (4) of unit (7) storage calculate first object storage time can are as follows: Calculate the difference of target reservoir storage and real-time reservoir storage, the quotient of calculating difference and specified water pumping amount, using the quotient as first object Storage time.

When the real-time reservoir storage miss the mark reservoir storage and the pump house (3) for corresponding to water storage box (4) in the real time data Wisdom water tank regulation unit (7) has turned on to when correspondence water storage box (4) water storage, illustrates that the reservoir storage in water storage box (4) not enough should The user of pump house (3) region uses, and wisdom water tank regulation unit (7) is needed to open corresponding municipal network water supply and water supply Conduit assembly is to corresponding water storage box (4) water storage；It is stored since wisdom water tank regulation unit (7) of the pump house (3) has turned on to corresponding Water tank (4) water storage, in order to improve the turnover rate of water body in corresponding water storage box (4), the reservoir storage in corresponding water storage box (4) reaches It can stop when target reservoir storage to corresponding water storage box (4) water storage.At this point, cloud platform big data analysis module (2) being capable of base The correspondence water storage box (4) of wisdom water tank regulation unit (7) storage of real-time reservoir storage, the pump house (3) in the real time data Maximum water requirement and target reservoir storage calculate the second target storage time, when reaching the second target storage time, cloud platform The wisdom water tank that big data analysis module (2) sends the second halt instruction to the pump house (3) regulates and controls unit (7), so that the pump house (3) wisdom water tank regulation unit (7) in stops corresponding municipal network water supply and water supply line component, to corresponding water storage box (4) water storage, wherein the mode for calculating the second target storage time is identical as the mode of first object storage time is calculated, herein It is not described further in detail.

When the real-time reservoir storage for corresponding to water storage box (4) in the real time data has reached target reservoir storage and the wisdom water tank tune Control unit (7) has turned on to when the correspondence water storage box (4) water storage, illustrates that the reservoir storage in the correspondence water storage box (4) has reached the pump The user in room (3) uses, it is no longer necessary to which wisdom water tank regulation unit (7) continues to keep it turned on to water storage box (4) water storage；This When, the wisdom water tank that cloud platform big data analysis module (2) can send third halt instruction to the pump house (3) regulates and controls unit (7), so that wisdom water tank regulation unit (7) of the pump house (3) closes to stop corresponding municipal network water supply and water supplying pipe Road component, to corresponding water storage box (4) water storage.

By the way that wisdom water tank regulation unit (7) in each pump house (3), hydraulic data monitoring unit (8), water quality prison is arranged Survey the accurate water storage that can obtain corresponding water storage box (4) in real time with processing unit (9), hydraulic pipeline data monitoring unit (10) Amount, water quality data and the real-time water consumption of user.Predict that each region corresponds to water storage based on target nerve network model simultaneously Target reservoir storage needed for case (4) is to get the watering law feature for arriving user；And according to the watering law of user spy Sign regulates and stores to water storage box (4) and specific mating water system facility is enabled to carry out water storage to corresponding water storage box (4), with Just before water use peak arriving, water is stored to the required target reservoir storage of each pump house (3) user in advance, is guaranteed with water height The pressure of water supply network will not be because of user, that is, hourly water consumption increase and by significant fluctuations when peak；Cloud platform can be passed through at this time Systems technology is managed as a whole in region, the Regulation capacity of all water storage boxes (4), to realize the tune of avoiding the peak hour to municipal network water supply It stores；By avoiding the peak hour, water supply of regulating and storing can shorten municipal network water supply and water supply line component pressure fluctuation range, municipal water supply pipe Net and the fluctuation of water supply line component pressure are small, while the pressure oscillation of municipal network water supply being avoided to increase water factory instantaneously for hydraulic pressure Power reduces the leakage loss of municipal network water supply and water supply line component, ensure that Water Works system even running, reduce water factory The reconstruction demand of municipal network water supply and water supply line component realizes country for the saving saving of water resource and auxiliary facility.

Further, since the real time data in water system of the present invention is very huge, and most of real time datas are that do not have The redundant data of standby reference value.In order to improve the accuracy of subsequent prediction target reservoir storage, cloud platform big data analysis module (2) it can also be used to determine that the real time data corresponds to water storage box (4) in the target nerve network model completed based on preparatory training Target reservoir storage before, the real time data is carried out after discrete cosine transform obtains de-redundancy based on discrete cosine transformation matrix Real time data.

The reality after discrete cosine transform obtains de-redundancy is wherein carried out to the real time data based on discrete cosine transformation matrix When data, comprising:

By the real-time reservoir storage for corresponding to water storage box (4) in real time data, active user water consumption, accumulate water supply, real-time The sample point data that water quality data and real-time water supply include is arranged in the first matrix, calculates the first matrix and discrete cosine transform The product of matrix obtains the second matrix, using each element in the second matrix as the real time data after de-redundancy.

Water system of the present invention carries out the specific of discrete cosine transform to the real time data based on discrete cosine transformation matrix The exemplary embodiment of embodiment is as follows: real time data includes multichannel data (being replaced with M circuit-switched data), is counted in real time in the present invention According to real-time reservoir storage, active user water consumption, accumulation water supply, real-time water quality data and the real-time water supply for including water storage box. At this point, M can be 5, it is assumed that every circuit-switched data includes N number of sampled point, then real time data is arranged in the first of N row M column Matrix A, the sample point data that a circuit-switched data is included form a column vector c of the first matrix A_j, for the first matrix A column to Measure c_j, the orthogonal matrix that discrete cosine transformation matrix U is N × N is defined, so that:

d_j=Uc_j, j=1,2 ..., N,

In above formula: d_jFor the second matrix column vector, U is discrete cosine transformation matrix, c_jFor the first matrix column vector, j For the line number of the second matrix, N is the second matrix column number, u_pqFor in discrete cosine transformation matrix U pth row q column element, P is the line number of discrete cosine transformation matrix, and q is the columns of discrete cosine transformation matrix.

After obtaining the second matrix, then using each element in the second matrix as the real time data after de-redundancy, as a result, De-redundancy is carried out to real time data based on discrete cosine transformation matrix, improves the accuracy of subsequent prediction target reservoir storage.

Due to normalization after data relative to the calculation amount of not normalized data small and convenience of calculation, right After real time data carries out de-redundancy, for subsequent prediction target reservoir storage convenience of calculation, additionally it is possible to the real-time number after de-redundancy According to being normalized, i.e., the mean value and standard of the real time data after cloud platform big data analysis module (2) calculating de-redundancy The real time data after being normalized is normalized to the real time data after de-redundancy based on mean value and standard deviation in difference, The target that real time data after the determining normalization of target nerve network model completed based on preparatory training corresponds to water storage box (4) stores Water.

Wherein specific cloud platform big data analysis module (2) calculates the mean value and standard of the real time data after de-redundancy The real time data after being normalized is normalized to the real time data after de-redundancy based on mean value and standard deviation in difference, Include:

Cloud platform big data analysis module (2) is specifically used for calculating characterization corresponding water storage box (4) in the second matrix and stores in real time The first mean value and the first standard deviation of first element of water, for each first element, calculate first element with it is described First difference of the first mean value calculates the first quotient of the first difference and the first standard deviation；

The second mean value and the second standard deviation that the second element of active user water consumption is characterized in the second matrix are calculated, it is right In each second element, second difference of the second element Yu second mean value is calculated, calculates the second difference and the second standard Second quotient of difference；

The third mean value and third standard deviation for calculating the third element of characterization accumulation water supply in the second matrix, for every A third element calculates the third difference of the third element Yu third mean value, calculates the third of third difference and third standard deviation Quotient；

The 4th mean value and the 4th standard deviation for calculating the fourth element that real-time water quality data is characterized in the second matrix, for Each fourth element, calculates the 4th difference of the fourth element Yu the 4th mean value, calculates the of the 4th difference and the 4th standard deviation Four quotients；

The 5th mean value and the 5th standard deviation for calculating the The Fifth Element that real-time water supply is characterized in the second matrix, for every A The Fifth Element calculates the 5th difference of the The Fifth Element and the 5th mean value, calculates the 5th of the 5th difference and the 5th standard deviation Quotient；

Using the first quotient, the second quotient, third quotient, the 4th quotient and the 5th quotient as the real-time number after normalization According to.

Wherein, the first element that real-time reservoir storage is characterized in the second matrix is normalized by following formula:

Wherein, X^*For the first quotient, X is the first element, and μ is the first mean value, and δ is the first standard deviation.

It is identical, place is normalized to second element, third element, fourth element and the The Fifth Element in the second matrix The formula of reason is identical as the first element, not described in detail herein.

As a result, in such a way that the real time data after de-redundancy is normalized, subsequent prediction target is improved The convenience of calculation of reservoir storage.

When water storage box (4) if in water quality it is poor not up to standard when, cloud platform big data analysis module (2) is not accused specifically Mating water system facility supplies water to user.At this point, in each pump house (3) in water storage box (4) corresponding with user's water supply, peace Installing is equipped with sterilizer and water tank self cleaning arrangement, and cloud platform big data analysis module (2) can be according in the real time data Real-time water quality data, determines whether the water quality in the pump house (3) in water storage box (4) corresponding with user's water supply reaches pre- bidding Standard, when the duration that the water quality in water storage box (4) is not up to preset standard reaches the first preset duration, cloud platform big data Analysis module (2) sends sterilizer open command, until disappearing in water storage box (4) corresponding with user's water supply in the pump house (3) Malicious device opens sterilizer and carries out disinfection processing to the water body in water storage box (4)；When the water quality in water storage box (4) is not up to default When the duration of standard reaches the second preset duration, cloud platform big data analysis module (2) sends water tank self cleaning arrangement and opens Instruction is opened, until the water tank self cleaning arrangement in the pump house (3) in water storage box (4) corresponding with user's water supply, opens water tank certainly Cleaning equipment carries out Cress processing to the water body in water storage box (4).Wherein, it is default to be less than described second for the first preset duration Duration.

Wherein, above-mentioned real-time water quality data is the digital expression of water quality situation in real time in corresponding water storage box (4), above-mentioned Determine whether the water quality in the correspondence water storage box (4) of the pump house reaches preset standard according to the real-time water quality data in the real time data The step of, comprising:

Judge whether real-time water quality data is less than predetermined quality data, if so, determining the correspondence water storage box of the pump house (3) (4) water quality in reaches preset standard, if not, determining that the water quality in the correspondence water storage box (4) of the pump house (3) is not up to default Standard.

When the duration that the water quality in the correspondence water storage box (4) of the pump house (3) is not up to preset standard reaches first in advance If when duration, illustrating that water quality is exceeded in short-term；At this point, cloud platform big data analysis module (2) sends sterilizer open command, until should Sterilizer in pump house (3) in water storage box (4) corresponding with user's water supply opens sterilizer to the water body in water storage box (4) Carry out disinfection processing；When the duration that water quality is not up to preset standard reaches the second preset duration, illustrate that water quality is long Shi Chaobiao, it is heavily contaminated, at this point, cloud platform big data analysis module (2) sends water tank self cleaning arrangement open command, Water tank self cleaning arrangement in the pump house (3) in water storage box (4) corresponding with user's water supply opens water tank self cleaning arrangement Cress processing is carried out to the water body in water storage box (4).Therefore, pass through the setting in cloud platform big data analysis module (2) The mode that sterilizer and self cleaning arrangement handle the water body in water storage box (4), can the water quality in water storage box (4) not When reaching preset standard, control sterilizer carries out disinfection or controls self cleaning arrangement to water storage box to the water body in water storage box (4) (4) water body in is cleaned, the water quality safety to be supplied water with the user ensured using water system of the present invention.

The wisdom water system in conclusion a kind of energy conservation based on cloud provided by the invention is avoided the peak hour, can pass through cloud Platform technology controls each wisdom water tank regulation unit and closes when real-time reservoir storage reaches target reservoir storage to be stored to water storage box (4) Water；It is managed as a whole in region by cloud platform technology, the Regulation capacity of all water storage boxes (4) realizes the mistake to water supply pipe net system Peak is regulated and stored water supply, shortens the range of pressure fluctuations of municipal network water supply and water supply line component by avoiding the peak hour to regulate and store to supply water, by It is small in the pressure oscillation of municipal network water supply and water supply line component, reduce the leakage of municipal network water supply and water supply line component Damage, ensure that water factory's municipal network water supply and water supply line component even running, reduces water factory's municipal network water supply and water supply Conduit assembly reconstructs demand, realizes country for the saving saving of water resource and auxiliary facility；Plan as a whole to manage by cloud platform technology The operation for managing each pump house (3) water storage box (4) corresponding sterilizer and self cleaning arrangement in region, can be in water storage box (4) When water quality is not up to preset standard, controls corresponding sterilizer and carry out disinfection or control automatically cleaning to the water body in water storage box (4) Equipment carries out Cress to the water body in water storage box (4), the water quality to be supplied water with the user guaranteed using water system of the present invention Safety.The present invention had both been able to achieve country for the saving saving of water resource and auxiliary facility, moreover it is possible to guarantee the water quality peace that user supplies water Entirely.

The skilled artisan will appreciate that: above-mentioned attached drawing is the schematic diagram of one embodiment, the mould in attached drawing Block or process not necessarily realize that all advantages of the invention are all necessary；Module in the present invention in corollary apparatus can be by It, also can be into while in one the device of the invention of integrated distribution according to the full content that embodiment in specific embodiment describes Row corresponding change is concentrated on respectively in two or more the device of the invention.The mould of embodiment i.e. in the specific embodiment of the invention Block can merge into a module, can also be further split into multiple submodule.

Above embodiments are only to illustrate the scheme that particular technique of the invention can be realized, and are not to its protection scope Limitation；Only referring to detailed description of the invention in above-described embodiment, so that those skilled in the art should manage It solves and realizes；It is other to technical solution illustrated in the above embodiments carry out modifications, or to part of technical characteristic into Capable equivalent replacement, these modify or replace the spirit and scope all without departing from the claimed technical solution of the present invention, It belongs to the scope of protection of the present invention.

Claims

Wisdom water system, including pump house, water storage box, water pump assembly, water supply line 1. a kind of energy conservation based on cloud is avoided the peak hour Component, water storage box are installed in pump house, and water storage box is connect by water supply line component with water pump assembly, pump house water storage box water pump machine Group has multiple, and installation settings has a water storage box in each pump house, and each water storage box is supplied water by water supply line component and water pump assembly User's connection, multiple pump houses inside water storage box are connect with water pump assembly by water supply line component, pass through city between multiple pump houses Political affairs water supply network connection, which is characterized in that further include data acquisition and transmission unit, pump house monitoring unit, the regulation of wisdom water tank Unit, hydraulic data monitoring unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit, cloud platform server and Cloud platform big data analysis module, wisdom water tank regulate and control the water quality monitoring of unit hydraulic data monitoring unit and processing unit pipe network water Force data monitoring unit and pump house, water storage box, water pump assembly, water supply line component are connected with each other, and wisdom water tank regulates and controls unit water The water quality monitoring of force data monitoring unit and processing unit hydraulic pipeline data monitoring unit, can be by the pump house acquired in real time, storage The waterpower water quality information data of water tank, water pump assembly, water supply line component, it is single via pump house monitoring unit and data acquisition transmission Member is transferred to Cloud Server, and instructs cloud platform big data analysis module according to Cloud Server, to more than preset standard waterpower water Pump house, water storage box, water pump assembly, the water supply line component of matter information data are adjusted, and make to be more than preset standard waterpower water quality Information data, which is adjusted to, to be complied with standard, and the cloud platform big data analysis module is embedded in cloud platform server internal, described Data acquisition and transmission unit, pump house monitoring unit, wisdom water tank regulation unit, hydraulic data monitoring unit, water quality monitoring and Processing unit, the quantity of hydraulic pipeline data monitoring unit are equal with the quantity of multiple pump house water storage box water pump assemblies, described every The pump house monitoring unit of a pump house can be used in sending and transmitting the regulation of the wisdom water tank in pump house unit, hydraulic data monitoring The monitoring data of unit, water quality monitoring and processing unit, hydraulic pipeline data monitoring unit, the pump house monitoring unit of each pump house Real time data is sent to cloud platform big data analysis module, the big number of cloud platform by the data acquisition of the pump house and transmission unit Determine that the real time data corresponds to the target storage of water storage box according to the target nerve network model that analysis module is completed based on preparatory training Water is then based on the real-time reservoir storage of corresponding water storage box, the operating state data of wisdom water tank regulation unit, supplies water in real time Data and the pipe network end hydraulic data that wisdom water tank regulates and controls unit, hydraulic pipeline data monitoring unit monitors are to determine It is no that water storage is carried out to water storage box；

The pump house monitoring unit of each pump house, can be used in the pump house hydraulic data monitoring unit send pump house to It supplies water and is monitored with the hydraulic data of the pressure of water, flow, real-time water supply, active user water usage data in family；

The pump house monitoring unit of each pump house can be used in the pipeline that the hydraulic pipeline data monitoring unit of the pump house is sent The hydraulic data monitoring of the pressure, flow, real-time water supply, real-time water-supply quantity data that supply water to pump house water storage box；

The pump house monitoring unit of each pump house can be used in water quality monitoring to the pump house and processing unit and water analysis The control of equipment water disinfection equipment；

The pump house monitoring unit of each pump house can be used in the wisdom water tank regulation unit analysis water storage box water to the pump house Matter is monitored in real time and is monitored to water storage box water quality, and the water quality data information of real-time monitoring is monitored list via the pump house Member and data acquisition transmission unit are transferred to cloud platform server and cloud platform big data analysis module, and refer to according to Cloud Server Cloud platform big data analysis module is enabled to carry out disinfection processing the water body for being more than preset standard water quality, the wisdom water tank regulation is singly The data information of first real-time monitoring further includes the real-time reservoir storage, active user water consumption and real-time water-supply quantity of water storage box；

The cloud that the cloud platform server and cloud platform big data analysis module use includes public cloud, private clound and mixing Cloud, the cloud can using distributed storage, redundant storage, cold and hot backup storage by the way of memory system data, the cloud skill Art is that cloud system can construct distributed computing by virtualization basic hardware resources, effectiveness calculates, load balancing calculates, parallel It calculates；

Wisdom water system is avoided the peak hour in the described energy conservation based on cloud can be by building one by multiple cloud platform servers Or the unified platform that storage server hardware is constituted, pump house, water storage box, water pump assembly, water supplying pipe are connected by local area network technology Road component, computer, corollary equipment carries out data communication and transmitting instruction in pump house；

Wisdom water system is avoided the peak hour in the described energy conservation based on cloud can be based on public communication network infrastructure, use IPsec VPN Virtual Private Network technology constructs the local area network based on public network, realizes the number of corollary equipment in cloud and each pump house According to transmission；

It is described should the wisdom water system of avoiding the peak hour of energy conservation based on cloud use PLC as pump house field controller collection site number According to receive control instruction, and data communication is carried out by Modbus industrial bus protocol specification and cloud；

It is described should the storage of pump house data in wisdom water system of avoiding the peak hour of energy conservation based on cloud realized by distributed data base, File process is handled by distributed file system and is realized；

It is described should energy conservation based on cloud avoid the peak hour that pass through building beyond the clouds compatible with public communication network for wisdom water system Local area network obtains pump house data and completes cloud storage, each with cloud platform big data analysis module analysis in cloud platform server end Pump house water data issues regulation and control instruction in turn；

The cloud platform big data analysis module, real-time water consumption and each pump house for wisdom water tank regulation unit users Data acquisition and the real time data that sends of transmission unit the real time data of each pump house is completed based on preparatory training Target nerve network model determines that the real time data corresponds to the target reservoir storage of water storage box, when corresponding to water storage box in the real time data Real-time reservoir storage, the not up to described target reservoir storage and wisdom water tank regulation unit is not when opening sends enabled instruction extremely The wisdom water tank regulates and controls unit, so that wisdom water tank regulation unit starts to give water storage box water storage, and based in the real time data Real-time reservoir storage, wisdom water tank regulation unit sends the highest reservoir storage of water storage box and the target reservoir storage calculates the One target storage time sends the first halt instruction to the wisdom water tank and regulates and controls when reaching the first object storage time Unit, with close wisdom water tank regulation unit, when the real-time reservoir storage in the real time data be not up to the target reservoir storage and When wisdom water tank regulation unit has turned on, unit is regulated and controled based on the real-time reservoir storage in the real time data, the wisdom water tank Specified water supply and the target reservoir storage calculate the second target storage time, when reaching the second target storage time When, it sends the second halt instruction to the wisdom water tank and regulates and controls unit, so that wisdom water tank regulation unit stops to the water storage box Water storage is stored when the real-time reservoir storage in the real time data has reached the target reservoir storage and has turned on wisdom water tank regulation unit It when water, sends third halt instruction to the wisdom water tank and regulates and controls unit, so that wisdom water tank regulation unit stops to the water storage Case water storage, the target nerve network model are real-time water consumption based on user, the real-time water consumption sample data of user and right It answers the sample reservoir storage of water storage box as model training data, initial target nerve network model is trained to obtain initial Neural network model, the target nerve network model are able to use the sample of family real-time water consumption sample data and corresponding water storage box This reservoir storage is associated, and the real-time water consumption sample data of user includes the real-time reservoir storage of sample of corresponding water storage box, sample Active user water consumption, sample cumulative water supply, the real-time water quality data of sample and the real-time water supply of sample.
The wisdom water system 2. a kind of energy conservation based on cloud according to claim 1 is avoided the peak hour, it is characterised in that described Cloud platform big data analysis module, additionally it is possible to for determining this in real time in the target nerve network model completed based on preparatory training Before data correspond to the target reservoir storage of water storage box, discrete cosine change is carried out to the real time data based on discrete cosine transformation matrix Real time data after getting de-redundancy in return, the mean value and standard deviation of the real time data after calculating the de-redundancy are described based on this Mean value and the standard deviation real time data after the de-redundancy are normalized the real-time number after being normalized According to the target nerve network model completed based on preparatory training determines that the real time data after the normalization corresponds to the mesh of water storage box Mark reservoir storage.
The wisdom water system 3. a kind of energy conservation based on cloud according to claim 2 is avoided the peak hour, it is characterised in that described Cloud platform big data analysis module specifically can be used in that real-time reservoir storage, the active user of water storage box will be corresponded in real time data The sample point data that water consumption, accumulation water supply, real-time water quality data and real-time water supply include is arranged in the first matrix, calculates The product of first matrix and discrete cosine transformation matrix obtains the second matrix, and each element in second matrix is made For the real time data for corresponding to pump house after de-redundancy.
The wisdom water system 4. a kind of energy conservation based on cloud according to claim 3 is avoided the peak hour, it is characterised in that described The cloud platform big data analysis module specifically can also be used to calculate first yuan that characterizes real-time reservoir storage in second matrix Element the first mean value and the first standard deviation, for each first element, calculate first element it is corresponding described first First difference of mean value, first difference of calculating and the first quotient of first standard deviation；

And the second mean value and the second standard deviation that the second element of active user water consumption is characterized in second matrix are calculated, For each second element, the second difference of corresponding second mean value of the second element is calculated, calculates described second Second quotient of difference and second standard deviation；

And the third mean value and third standard deviation of the third element of characterization accumulation water supply in second matrix are calculated, for Each third element calculates the third difference of the corresponding third mean value of the third element, calculates the third difference With the third quotient of the third standard deviation；

And the 4th mean value and the 4th standard deviation for the fourth element that real-time water quality data is characterized in second matrix are calculated, it is right In each fourth element, the 4th difference of corresponding the 4th mean value of the fourth element is calculated, it is poor to calculate the described 4th 4th quotient of value and the 4th standard deviation；

And the 5th mean value and the 5th standard deviation that the The Fifth Element of real-time water supply is characterized in second matrix are calculated, for Each The Fifth Element calculates the 5th difference of corresponding the 5th mean value of the The Fifth Element, calculates the 5th difference With the 5th quotient of the 5th standard deviation；

Using first quotient, second quotient, the third quotient, the 4th quotient and the 5th quotient as institute Real time data after stating normalization.
The wisdom water system 5. a kind of energy conservation based on cloud according to claim 1 is avoided the peak hour, it is characterised in that described The training process of target nerve network model, specifically:

The sample reservoir storage of the real-time water consumption sample data of user and corresponding water storage box that training uses is obtained, the user is real-time Water consumption sample data includes that the real-time reservoir storage of sample, sample active user water consumption, sample cumulative of corresponding water storage box supply water Amount, the real-time water quality data of sample and the real-time water supply of sample；

The sample reservoir storage of the real-time water consumption sample data of the user and corresponding water storage box is inputted into initial neural network model In, the initial neural network model includes input layer, hidden layer, output layer and establishes between input layer and output layer Equation group relationship, the equation group relationship of the foundation between input layer and output layer is excitation function；

According to the real-time water consumption sample data of the user and the sample reservoir storage of corresponding water storage box, each node of the input layer with Between weight coefficient, each node of the hidden layer between each node of hidden layer and the output node layer of the output layer The equation group relationship of weight coefficient and the foundation between the input layer and the output layer；

Solve the equation group, obtain the weight coefficient between each node of the input layer and each node of the hidden layer value, The value of weight coefficient between each node of hidden layer and the output node layer of the output layer；

By the value of the weight coefficient between each node of the input layer and each node of the hidden layer, each node of the hidden layer with Described in the value and the real-time water consumption sample data of the user of weight coefficient between the output node layer of the output layer substitute into In equation group, the initial reservoir storage that the real-time water consumption sample data of the user corresponds to water storage box is solved；

Calculate the difference value between the initial reservoir storage of the corresponding water storage box and the sample reservoir storage of the corresponding water storage box；

When the difference value is greater than default discrepancy threshold, first weight coefficient and described the are adjusted based on the difference value Two weight coefficients are returned and are executed the sample reservoir storage input of the real-time water consumption sample data of the user and corresponding water storage box just Step in beginning neural network model, first weight coefficient are first input layer and the first hidden layer section The first weight coefficient between point, second weight coefficient are second input layer and the described first hiding node layer Between the second weight coefficient；

When the difference value is not more than default discrepancy threshold, training is completed, is obtained comprising the real-time water consumption sample data of user With the target nerve network model of the corresponding relationship of the sample reservoir storage of the corresponding water storage box.
The wisdom water system 6. a kind of energy conservation based on cloud according to claim 5 is avoided the peak hour, it is characterised in that described Input layer includes the first input layer, the second input layer and third input layer, and the hidden layer includes first hidden It hides node layer, the second hiding node layer and third and hides node layer, the output layer includes output node layer, described according to The real-time water consumption sample data of user and the sample reservoir storage of corresponding water storage box, each node of the input layer and the hidden layer are each Weight coefficient and institute between weight coefficient, each node of the hidden layer between node and the output node layer of the output layer The step of stating the equation group relationship established between the input layer and the output layer, comprising:

According between the real-time water consumption sample data of the user, first input layer and the first hiding node layer The first weight coefficient, the second weight coefficient between second input layer and the first hiding node layer and described The first constant between third input layer and the first hiding node layer establishes the first linear equation；

According between the real-time water consumption sample data of the user, first input layer and the second hiding node layer Third weight coefficient, the 4th weight coefficient between second input layer and the second hiding node layer and described The second constant between third input layer and the second hiding node layer establishes the second linear equation；

It is hidden between node layer according to the real-time water consumption sample data of the user, first input layer and the third The 5th weight coefficient, second input layer and the third hide the 6th weight coefficient between node layer and described The third constant that third input layer and the third are hidden between node layer establishes third linear equation；

It is excitation function, user reality according to equation group relationship of the foundation between the input layer and the output layer Hourly water consumption sample data corresponds to the sample reservoir storage of water storage box, the functional value of first linear equation, described second linearly Between the functional value of the functional value of equation and the third linear equation, the first hiding node layer and the output node layer The 7th weight coefficient, the 8th weight coefficient and described the between the second hiding node layer and the output node layer The 9th weight coefficient between three hiding node layers and the output node layer establishes the 4th functional equation；

First linear equation described in simultaneous, second linear equation, the third linear equation and the 4th functional equation, Obtain the equation group between the input layer and the output layer；

It is described to solve the equation group, obtain the weight coefficient between each node of the input layer and each node of the hidden layer The step of value of weight coefficient between the output node layer of value, each node of the hidden layer and the output layer, comprising:

The equation group is solved, the value, the value of second weight, the value of the third weight, institute of first weight are obtained State the value of the 4th weight, the value of the 5th weight, the value of the 6th weight, the value of the 7th weight, the 8th power The value of the value of weight and the 9th weight；

It is described respectively to save the value of the weight coefficient between each node of the input layer and each node of the hidden layer, the hidden layer The value and the real-time water consumption sample data of the user of weight coefficient between point and the output node layer of the output layer substitute into In the equation group, the step of real-time water consumption sample data of the user corresponds to the initial reservoir storage of water storage box is solved, is wrapped It includes:

By the value of first weight, the value of second weight, the value of the third weight, the value of the 4th weight, institute State the value of the 5th weight, the value of the 6th weight, the value of the 7th weight, the value of the 8th weight and the 9th power The real-time water consumption sample data of the value and the user of weight substitutes into the equation group, solves the real-time water consumption sample of the user Notebook data corresponds to the initial reservoir storage of water storage box.
The wisdom water system 7. a kind of energy conservation based on cloud according to claim 5 is avoided the peak hour, it is characterised in that described The functional value of excitation function are as follows: the sum of drive factor and target max, the target max are that the user uses water in real time The maximum value among sample data and target product is measured, the target product is that the drive factor is used in real time with the user The product of water sample data.
The wisdom water system 8. a kind of energy conservation based on cloud according to claim 1 is avoided the peak hour, it is characterised in that described Also installation settings has sterilizer and water storage box self cleaning arrangement, the cloud platform big data analysis mould in the water storage box of each pump house Block, additionally it is possible to for determining whether the water quality in the water storage box of the pump house reaches according to the real-time water quality data in the real time data Preset standard sends sterilizer open command when the duration that water quality is not up to preset standard reaches the first preset duration Sterilizer to the pump house water storage box, so that the sterilizer is opened, when the duration that water quality is not up to preset standard reaches When to the second preset duration, the water storage box automatically cleaning sent in water storage box self cleaning arrangement open command to the pump house water storage box is set Standby, so that the water storage box self cleaning arrangement is opened, first preset duration is less than second preset duration.
The wisdom water system 9. a kind of energy conservation based on cloud according to claim 8 is avoided the peak hour, it is characterised in that described Real-time water quality data is real-time water quality, and the real-time water quality data according in the real time data determines in the water storage box of the pump house Water quality the step of whether reaching preset standard, comprising:

Judge whether the real-time water quality is less than predetermined quality, if so, determining that the water quality in the water storage box of the pump house reaches pre- Bidding is quasi-, if not, determining that the water storage water quality of the pump house is not up to preset standard.