CN114696466A - State monitoring system and method suitable for low-voltage transformer area power distribution - Google Patents
State monitoring system and method suitable for low-voltage transformer area power distribution Download PDFInfo
- Publication number
- CN114696466A CN114696466A CN202210477862.8A CN202210477862A CN114696466A CN 114696466 A CN114696466 A CN 114696466A CN 202210477862 A CN202210477862 A CN 202210477862A CN 114696466 A CN114696466 A CN 114696466A
- Authority
- CN
- China
- Prior art keywords
- box
- meter box
- monitoring terminal
- station area
- meter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H02J13/12—
-
- H02J13/36—
-
- H02J3/00142—
Landscapes
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
本发明涉及一种适用于低压台区配电的状态监测系统及方法。其包括集中器以及与所述集中器采用基于HPLC网络通信的台区配电监测装置；集中器与变压器出线侧监控终端、分支箱监控终端以及表箱监控终端基于HPLC网络通信后，集中器确定所述集中器所在台区的配电拓扑关系，所确定的配电拓扑关系包括台区户变关系、台区分支拓扑关系以及台区表箱拓扑关系，其中，集中器进行台区表箱拓扑关系识别时，采用基于CPSO的K‑means集群分类算法确定任一表箱监控终端所对应台区表箱内的用户表数量，并通过所述算法的分类结果得出户表与表箱之间的对应关系。本发明能有效实现对低压台区配电的状态监测，提高低压台区配电监测的智能化程度。
The invention relates to a state monitoring system and method suitable for power distribution in low-voltage station areas. It includes a concentrator and a station area power distribution monitoring device that communicates with the concentrator based on an HPLC network; after the concentrator communicates with the transformer outlet side monitoring terminal, the branch box monitoring terminal and the meter box monitoring terminal based on the HPLC network, the concentrator determines. The power distribution topology relationship of the station area where the concentrator is located, the determined power distribution topology relationship includes the station area household change relationship, the station area branch topology relationship, and the station area meter box topology relationship, wherein the concentrator performs the station area meter box topology relationship. During relationship identification, the K-means cluster classification algorithm based on CPSO is used to determine the number of user tables in the meter box of the station area corresponding to any meter box monitoring terminal, and the classification result of the algorithm is used to obtain the difference between the user meter and the meter box. corresponding relationship. The invention can effectively realize the state monitoring of the power distribution in the low-voltage station area, and improve the intelligence level of the power distribution monitoring in the low-voltage station area.
Description
Technical Field
The invention relates to a state monitoring system and a state monitoring method, in particular to a state monitoring system and a state monitoring method suitable for low-voltage distribution area power distribution.
Background
At present, two modes of on-site manual detection and on-line intelligent monitoring are mainly adopted for identifying the topological relation of the power distribution network in the low-voltage distribution area.
For a topology identification mode adopting field manual detection, the manual detection requires a worker to carry out field measurement by using a platform area topology identification device, the method has low identification precision, and particularly when the load of a power grid user changes, the fault positioning is difficult due to untimely identification. For a topology identification mode adopting online intelligent monitoring, methods such as carrier communication and pulse current based on a monitoring terminal and an HPLC communication unit hardware module are mostly adopted at present, the identification accuracy is high, but the pulse current needs to be controlled within a certain amplitude range, the problems of 'station area crosstalk' and the like exist in the carrier communication, the precision is easily influenced, certain difficulty also exists in operation, and a better improvement scheme is not obtained at present.
With the continuous expansion of the user scale of the intelligent electric energy meter, the big data topology identification method based on the intelligent electric energy meter monitoring terminal has more obvious advantages in identification accuracy and identification efficiency. However, how to effectively realize the topology identification of the low-voltage distribution area is still a technical problem to be solved continuously.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a state monitoring system and method suitable for low-voltage distribution area power distribution, which can effectively confirm the number of user meters in a meter box of the distribution area, effectively realize the state monitoring of the low-voltage distribution area power distribution and improve the intelligent degree of the low-voltage distribution area power distribution monitoring.
According to the technical scheme provided by the invention, the state monitoring system suitable for low-voltage distribution area comprises a concentrator and a distribution area distribution monitoring device which is communicated with the concentrator on the basis of an HPLC network; the distribution monitoring device in the distribution area comprises a transformer outgoing side monitoring terminal, a plurality of branch box monitoring terminals and a plurality of meter box monitoring terminals, wherein the transformer outgoing side monitoring terminals are in adaptive connection with the transformer outgoing end of the distribution area;
after the concentrator communicates with a monitoring terminal at the outgoing line side of a transformer, a branch box monitoring terminal and a meter box monitoring terminal on the basis of an HPLC network, the concentrator determines the distribution topological relation of a transformer area where the concentrator is located, and the determined distribution topological relation comprises a transformer area household variation relation, a transformer area branch topological relation and a transformer area meter box topological relation, wherein when the concentrator determines the topology of the meter box in the transformer area, the number of user meters in the meter box in the transformer area corresponding to any meter box monitoring terminal is determined by adopting a CPSO-based K-means cluster classification algorithm, and the corresponding relation between the user meters and the meter box in the transformer area is obtained through the classification result of the CPSO-based K-means cluster classification algorithm.
When the CPSO-based K-means cluster classification algorithm is adopted to determine the number of user meters in the district meter box corresponding to the meter box monitoring terminal, the method comprises the following steps:
the method comprises the following steps that 1, N household meter working parameter samples of N household meters in a monitored area meter box are obtained through a meter box monitoring terminal, wherein each household meter working parameter sample comprises three-phase working voltage, three-phase working current and three-phase active power which are sampled at the same sampling time point;
step 2, initializing a chaotic particle swarm algorithm and setting the clustering number K of a K-means clustering method according to the N user table working parameter samples, and randomly selecting K user table working parameter samples from the N user table working parameter samples as the initial position of a particle;
step 3, updating the current speed and position of each particle in the population; calculating the target function fitness value and the global optimal position of the population after the speed and the position are updated;
step 4, comparing the obtained global optimal position with a preset chaotic optimization condition, and skipping to step 5 when the obtained global optimal position is matched with the chaotic optimization condition meeting the preset chaotic optimization condition, or skipping to step 3;
step 5, generating a new population according to the chaotic particle swarm algorithm, comparing the generated new population with the termination condition of the chaotic particle swarm algorithm, and skipping to step 6 when the generated new population is matched with the termination condition of the chaotic particle swarm algorithm, or skipping to step 3;
step 6, determining the global optimal position of the chaotic particle swarm algorithm, and performing K-means cluster analysis on the N user table working parameters based on the cluster number K by taking the determined global optimal position of the particles as a cluster center;
step 7, after the K-means cluster analysis, comparing the state of the cluster analysis with a cluster termination condition, and jumping to step 9 when the state of the cluster analysis is matched with the cluster termination condition, or jumping to step 8;
step 8, changing k to k +1 to update the clustering number, and jumping to the step 6;
step 9, outputting a result of K-means cluster analysis based on the cluster number K, and obtaining the number of user meters in the distribution area meter box consistent with the cluster number K according to the determined cluster number K;
and step 10, determining a distribution area meter box to which any user meter belongs by combining the user meter ID addresses in the user meter working parameter samples in the clustering classification results, and obtaining the topological relation between the distribution area meter box and the user meters.
And the clustering termination condition is that the maximum iteration times are reached, or the effective value of the target function calculated in the clustering process is matched with a preset threshold value of the effective value of the target function.
The meter box monitoring terminal modulates the address of each user meter into the three-phase working current, and when the number of the user meters in the meter box of the transformer area is obtained, the corresponding address of the user meter in the meter box of the transformer area is obtained at the same time.
The concentrator determines the station area indoor variation relationship by adopting a power frequency distortion signal mode;
and the transformer outgoing line side monitoring terminal, the branch box monitoring terminal, the meter box monitoring terminal and the concentrator determine the distribution area branch topology in a pulse current mode.
A risk early warning model based on a deep convolutional neural network Alexnet is configured in a monitoring terminal at the outlet side of the transformer, a branch box monitoring terminal and/or a meter box monitoring terminal,
for a transformer outgoing line side monitoring terminal, processing the temperature of a wiring terminal of the transformer outgoing line side, the three-phase current value of the transformer outgoing line side and the three-phase leakage current of the transformer outgoing line side through a risk early warning model to determine a risk early warning value of the transformer outgoing line side;
processing the branch box wiring terminal temperature, the branch box three-phase current value and the branch box three-phase leakage current through a risk early warning model configured in the branch box monitoring terminal so as to determine a risk early warning value of the branch box;
and (4) a meter box monitoring terminal is monitored, and the temperature of a wiring terminal of the meter box in the area, the three-phase current value of the meter box in the area and the three-phase leakage current of the meter box in the area are processed through a risk early warning model configured in the meter box monitoring terminal so as to determine the risk early warning value of the meter box in the area.
When a risk early warning model is constructed, the method comprises the following steps:
step S1, providing a neural network basic model based on a deep convolutional neural network Alexnet, and configuring model basic state parameters of the neural network basic model, wherein the configured model basic state parameters comprise precision control parameters and a learning rate;
s2, making a sample data set for training a basic model of the neural network, wherein the sample data set comprises working state parameters under various working states, the working states comprise a normal working state, a full load working state and a fault state, the working state parameters comprise working process parameters and early warning state parameters corresponding to the working process parameters, and the working process parameters comprise temperature, three-phase current and three-phase leakage current;
dividing the manufactured sample data set into a required training sample set and a required verification sample set, wherein the training sample set is used for training the basic neural network model, and the verification sample set is used for verifying the trained basic neural network model;
step S3, training the neural network basic model by using a training sample set, and verifying the trained neural network basic model by using a verification sample set when a training termination condition is reached;
and when the verification of the neural network basic model by using the verification sample set is matched with the verification conditions, obtaining a risk early warning model based on the deep convolution neural network Alexnet by using the trained neural network basic model.
In step S2, the sample data set includes a plurality of sample data obtained by processing the sampling basic data, where the processing of the sampling basic data includes feature value extraction, oscillation mode screening, and preprocessing performed in sequence.
The meter box monitoring terminal is characterized by further comprising an environment monitoring device which is electrically connected with the meter box monitoring terminal in an adaptive mode, the environment state parameters of the environment where the meter box monitoring terminal is located can be monitored by the environment monitoring device, and the monitored environment state parameters are transmitted to the concentrator.
A state monitoring method suitable for low-voltage distribution area power distribution utilizes the state monitoring system to carry out required state monitoring on the low-voltage distribution area.
The invention has the advantages that: determining the number of user meters in the district meter box corresponding to any meter box monitoring terminal by adopting a K-means cluster classification algorithm based on CPSO (compact peripheral component analysis) to effectively confirm the number of the user meters in the district meter box; the risk early warning model based on the deep convolutional neural network Alexnet can effectively monitor the state of low-voltage distribution area power distribution and improve the intelligent degree of low-voltage distribution area power distribution monitoring.
Drawings
FIG. 1 is a schematic diagram of a specific embodiment of a low pressure zone.
Fig. 2 is a schematic diagram of a specific topological relationship of the low-voltage transformer area.
FIG. 3 is a flowchart of determining the number of user meters in a distribution area meter box corresponding to any meter box monitoring terminal based on the CPSO K-means cluster classification algorithm.
Fig. 4 is a schematic diagram of constructing a risk early warning model according to the present invention.
Detailed Description
The invention is further illustrated by the following specific figures and examples.
In order to effectively realize the state monitoring of the low-voltage distribution area and improve the intelligent degree of the low-voltage distribution area distribution monitoring, the low-voltage distribution area distribution monitoring system comprises a concentrator and a distribution area distribution monitoring device which is communicated with the concentrator on the basis of an HPLC network; the distribution monitoring device in the distribution area comprises a transformer outgoing side monitoring terminal, a plurality of branch box monitoring terminals and a plurality of meter box monitoring terminals, wherein the transformer outgoing side monitoring terminals are in adaptive connection with the transformer outgoing end of the distribution area;
after the concentrator communicates with a transformer outgoing line side monitoring terminal, a branch box monitoring terminal and a meter box monitoring terminal on the basis of an HPLC network, the concentrator determines a distribution topology state of a transformer area where the concentrator is located, the determined distribution topology state comprises a transformer area household variable relation, a transformer area branch topology and a transformer area meter box topology, and when the concentrator determines the transformer area meter box topology, the number of user meters in a transformer area meter box corresponding to any meter box monitoring terminal is determined by adopting a CPSO-based K-means cluster classification algorithm.
Specifically, the concentrator can adopt the existing commonly used form, and the concentration district needs to possess the ability of HPLC communication, and the HPLC network of district distribution monitoring is established with district distribution monitoring devices through HPLC communication interactive mode to the concentrator promptly, therefore, district distribution monitoring devices also need to possess the ability of HPLC communication, and the mode that concentrator, district distribution monitoring devices specifically realized HPLC communication ability can be selected according to actual need to can satisfy based on HPLC network communication as the standard, and here is no longer specifically explained.
During specific implementation, the monitoring terminal on the outlet side of the transformer can monitor the power supply condition of the outlet end of the transformer in the transformer area, the monitoring terminal of the branch box can monitor the working state of power distribution of the branch control electric box, and the monitoring terminal of the meter box can monitor the working state of power distribution of the meter box in the transformer area; fig. 1 is a specific connection structure diagram of current low-voltage transformer district, and in fig. 1, transformer outgoing line side monitor terminal and the outgoing line side adaptation of transformer district are connected, and a plurality of branch distribution boxes and transformer outgoing line side monitor terminal adaptation are connected, and every branch distribution box can be connected with one or more district table case adaptations, and the distribution form that transformer outgoing line side monitor terminal concrete cooperation between formed low-voltage transformer district is unanimous with current.
During specific implementation, the transformer outgoing line side monitoring terminal can be an intelligent molded case circuit breaker, and the intelligent molded case circuit breaker can adopt the existing common form so as to meet the requirement of HPLC network communication and monitor the power supply of the transformer in the transformer area. When the transformer outlet side monitoring terminal adopts an intelligent molded case circuit breaker, and when the current flowing through the molded case circuit breaker exceeds a set value, the transformer outlet side monitoring terminal can disconnect the power supply of a branch distribution box connected to the transformer outlet side monitoring terminal. Every branch block terminal and a branch case monitor terminal adaptation connection, every platform district table case and a table case monitor terminal adaptation connection, branch case monitor terminal, table case monitor terminal's specific form can be selected as required to can satisfy HPLC network communication, and can realize that required distribution work control is accurate, and here is no longer repeated.
In fig. 2, a specific topology schematic diagram of a low-voltage transformer area is shown, where a concentrator is used as a master node, a monitoring terminal on a transformer outgoing line side directly connected to the concentrator is a primary branch, a branch box monitoring terminal adaptively connected to the monitoring terminal on the transformer outgoing line side is a secondary branch, a meter box monitoring terminal adaptively connected to the branch box monitoring terminal is a meter box level of the transformer area, and the condition of a specific topology relationship is consistent with that of the existing meter box level, and is not repeated here.
In the embodiment of the invention, the concentrator can be used for realizing the identification and confirmation of the distribution topological state of the whole low-voltage transformer area, and the distribution topological state determined by the concentrator area comprises transformer area indoor transformation relation, transformer area branch topology and transformer area meter box topology. In specific implementation, when the transformer-substation relationship is identified, the concentrator determines the transformer-substation relationship by using a power frequency distortion signal, wherein when the transformer-substation relationship is determined, the concentrator transmits a power frequency distortion signal through an HPLC network, in the whole low-voltage transformer substation, the outgoing line side monitoring terminal, the branch box monitoring terminal and/or the meter box monitoring terminal of the transformer receiving the power frequency distortion signal feed back a power frequency distortion receiving confirmation message to the concentrator, the concentrator obtains the transformer-substation relationship according to the fed back power frequency distortion receiving confirmation message, and according to the obtained transformer-substation relationship, the concentrator can confirm the specific condition of the distribution monitoring device in the transformer-substation when the whole low-voltage transformer substation is distributed, such as the corresponding quantity relationships among the outgoing line side monitoring terminal, the branch box monitoring terminal and the meter box monitoring terminal in the transformer-substation distribution monitoring device, and the specific condition of the transformer-substation relationship is consistent with the existing condition, the situation of determining the station area diversity relationship by using the power frequency distortion signal is also consistent in the prior art, and is well known by those skilled in the art, and is not described herein again.
After the transformer area household transformation relation is determined, the transformer outgoing line side monitoring terminal, the branch box monitoring terminal, the meter box monitoring terminal and the concentrator determine the transformer area branch topology in a pulse current mode, and the condition that the transformer area branch topology is determined in the pulse current mode is consistent with the existing condition. After the distribution area branch topology is determined, the concentrator can determine the specific conditions of the first-stage branch, the second-stage branch and the distribution area meter box-stage topology in the low-voltage distribution area. After the first-stage branch, the second-stage branch and the distribution area meter box-stage topology in the low-voltage distribution area topology are obtained, the number of the user meters in the distribution area meter box monitored by each meter box monitoring terminal and the corresponding relation between the user meters and the corresponding meter boxes cannot be determined.
In order to determine the number of user meters in a meter box of a station area monitored by any meter box monitoring terminal and the correspondence between the user meters and the meter box, the embodiment of the invention adopts a K-means cluster classification algorithm based on a CPSO (chaotic particle swarm optimization) to determine the number of the user meters in the meter box of the station area monitored by any meter box monitoring terminal and the correspondence between the user meters and the meter box. In specific implementation, the K-means cluster classification algorithm based on the CPSO has the basic idea that: aiming at the problem that the traditional K-means clustering algorithm is sensitive to the value of an initial clustering center, the improved chaotic particle swarm algorithm is used for optimizing the position of the clustering center in the K-means clustering algorithm, the influence of the initial clustering center and the possibility of falling into a local optimal solution are reduced, clustering division under different clustering numbers is optimized, accordingly, the optimal classification result of the table area meter box family table is obtained, and the unique ID address issued by a concentrator to each family table is combined, so that meter box-family table relation identification can be realized.
In the embodiment of the invention, when a concentrator determines the topology of a transformer area meter box, the quantity of user meters in the transformer area meter box corresponding to any meter box monitoring terminal and the corresponding relation between the user meters and the meter box are determined by adopting a K-means cluster classification algorithm based on CPSO; specifically, for each meter box monitoring terminal, the concentrator determines the number of user meters in the meter box of the distribution area corresponding to the corresponding meter box monitoring terminal and the corresponding relation between the user meters and the meter box to which the user meters belong by adopting a K-means cluster classification algorithm based on CPSO. The following describes a process of determining the number of user meters in the table box of the distribution area corresponding to the corresponding table box monitoring terminal based on the CPSO K-means cluster classification algorithm.
As shown in fig. 3, when determining the number of user meters in the table box of the distribution area corresponding to the monitoring terminal of the table box by using the K-means cluster classification algorithm based on the CPSO, the method includes the following steps:
the method comprises the following steps that 1, N household meter working parameter samples of N household meters in a monitored area meter box are obtained through a meter box monitoring terminal, wherein each household meter working parameter sample comprises three-phase working voltage, three-phase working current and three-phase active power which are sampled at the same sampling time point;
specifically, for N working parameter samples of N user meters, actual demand sampling can be performed, for example, for a meter box monitoring terminal, 48-hour three-phase working voltage, three-phase working current and three-phase active power data (with consistent time) of all the user meters in a corresponding area meter box are selected to form a sample sequence, the sampling interval can be 15 minutes, and 192 data points are acquired in 48 hours; for the power supply state of three-phase alternating current, the dimension of any user table working parameter sample is 3 at the moment. And for any data point, the corresponding user table working parameter samples of all the user tables are included.
Therefore, during specific implementation, the meter box monitoring terminal needs to have the capability of sampling the voltage, the current and the active power of the user meter in the corresponding area meter box besides the HPLC communication capability, and the meter box monitoring terminal can specifically adopt the technical means commonly used in the technical field to sample the voltage, the current and the active power of the user meter, which is well known by those skilled in the art and is not repeated here.
Step 2, initializing a chaotic particle swarm algorithm and setting the clustering number K of a K-means clustering method according to the N user table working parameter samples, and randomly selecting K user table working parameter samples from the N user table working parameter samples as the initial position of a particle;
repeating the generation process of the initial particle positions to generate a population P with M particles, wherein after the particle swarm algorithm is initialized, the cluster number k is smaller than the number N of the working parameter samples of the user table;
specifically, after obtaining N user table working parameter samples, providing a chaotic particle swarm algorithm and performing corresponding initialization according to the N user table working parameter samples, generally, after initializing the chaotic particle swarm algorithm, obtaining a population scale P, wherein the initial position of each particle can be determined by randomly selecting k user table working parameter samples.
Of course, at initialization, the velocity of each particle needs to be initialized, and the initial velocity of each particle may be randomly generated or 0. The inertia weight coefficient w, the learning factor c1, and the learning factor c2 may be initialized or configured according to a conventional method, for example, the values of the learning factor c1 and the learning factor c2 may be c 1-c 2-2.05, which is well known in the art and will not be described herein again. Further, for the random number r1 and the random number r2, the random number r1 and the random number r2 may be random numbers within [0, 1 ]. Specifically, the method and process for initializing the chaotic particle swarm algorithm are consistent with those in the prior art, which are well known to those skilled in the art and will not be described herein again.
In addition, according to N user table working parameter samples N, the clustering number K of the K-means method is given, and specifically, the clustering number K is smaller than the number N of the user table working parameter samples.
Step 3, updating the current speed and position of each particle in the population; calculating the target function fitness value and the global optimal position of the population after the speed and the position are updated;
specifically, the population may be the initialized population described above, or a population generated by subsequent updating. For the population, the current speed and position of each particle in the particle swarm are updated by adopting the common technical means in the technical field, and after the current speed and position of each particle are updated, the target function fitness value and the global optimal position of the population are calculated by adopting the common technical means in the technical field. Therefore, the specific processes of updating the speed and the position of the particle, and calculating the fitness value of the objective function and the global optimal position of the population are well known to those skilled in the art, and will not be described herein again.
Step 4, comparing the obtained global optimal position with a preset chaotic optimization condition, and skipping to step 5 when the obtained global optimal position is matched with the chaotic optimization condition meeting the preset chaotic optimization condition, or skipping to step 3;
specifically, the preset chaotic optimization condition comprises an average value of the global optimal position change rate and the optimization times; when the current execution times are less than the preset optimization times and the change rate of the global optimal position is determined to be lower than the average value of the change rate of the global optimal position according to the obtained global optimal position, performing chaotic optimization, namely jumping to the step 5, otherwise, judging that the chaotic optimization condition is not met, and jumping to the step 3;
in the embodiment of the invention, the change rate of the global optimal position can be generally obtained by the change of two adjacent global optimal positions, and is specifically consistent with the prior art. During chaotic optimization, the number of times of optimization execution generally needs to be counted, and the number of times of chaotic optimization needs to be not higher than the preset number of times of optimization, and during specific implementation, the number of times of optimal execution is the number of times of comparison between the global optimal position and the preset chaotic optimization condition in the current step 4. The average value of the global optimal position change rate and the specific condition of the number of optimization times may be selected and determined according to actual needs, which are well known to those skilled in the art and will not be described herein again.
In specific implementation, the chaos optimization condition is judged not to be satisfied, specifically, the global optimal position change rate is greater than or equal to the average value of the preset global optimal position change rates.
Step 5, generating a new population according to the chaotic particle swarm algorithm, comparing the generated new population with the termination condition of the chaotic particle swarm algorithm, and skipping to step 6 when the generated new population is matched with the termination condition of the chaotic particle swarm algorithm, or skipping to step 3;
specifically, a new population is generated by adopting a common technical means in the technical field, for example, when the new population is generated: updating the speed and the position of 95% of winning individuals with higher fitness values, and recording the obtained population as a sub-population Popull; and mapping 5% of non-dominant individuals with lower fitness values in the population as chaotic variables, carrying out required variation on the obtained chaotic variables to obtain a varied sub-population Popul2, and combining the varied sub-population Popul2 and the dominant population Popul1 into a new population Pn, namely generating a new population.
The method specifically refers to that the number of times of optimization reaches a preset number of optimization or the global optimal position change rate is matched with a preset average value of the global optimal position change rate, the matching specifically refers to that the global optimal position change rate is consistent with the preset average value of the global optimal position change rate or the difference value between the global optimal position change rate and the preset average value of the global optimal position change rate is within an allowed range, and the allowed range of the difference value can be selected and determined according to actual conditions, which is well known by persons skilled in the art, and is not described herein again.
Step 6, determining the global optimal position of the chaotic particle swarm algorithm, and performing K-means cluster analysis on the N user table working parameters based on the cluster number K by taking the determined global optimal position of the particles as a cluster center;
specifically, after the objective function fitness value and the global optimal position of the population are obtained, the global optimal position is configured as a clustering center, K-means clustering analysis is performed based on the clustering number K, and after the clustering number K and the clustering center are determined, K-means clustering is performed on N user table working parameter samples specifically, and the specific clustering mode and process are consistent with those of the prior art, which are well known to those skilled in the art and are not described herein again. Step 7, after the K-means cluster analysis, comparing the state of the cluster analysis with a cluster termination condition, and jumping to step 9 when the state of the cluster analysis is matched with the cluster termination condition, or jumping to step 8;
specifically, the clustering termination condition may be that the maximum iteration number is reached, or that the objective function effective value calculated in the clustering process is matched with a preset objective function effective value threshold, where the maximum iteration number may be determined according to actual needs, and the objective function effective value threshold may also be determined according to actual monitoring requirements, and the like, which are well known to those skilled in the art and are not described herein again.
In the clustering process, the calculation of the target function effective value can be realized by adopting the technical means commonly used in the technical field, for example, the target function effective value can be obtained by calculating through the following formula, specifically:
in general, the effective value Gk of the objective function is:
wherein x represents a user table working parameter sample; k is the number of clusters; e.g. of the typeiIs a cluster EiThe cluster center of (a); n isiAnd the number of the working parameter samples of the user table in the i clusters. For the clustering center, namely the global optimal position output by the chaotic particle swarm algorithm, the number of the working parameter samples of the user table in the i clusters can be directly determined and obtained, and is specifically consistent with the existing methodAs is well known to those skilled in the art. dist is a sample of the working parameters of the calculation user table and the clustering center eiThe distance between them.
Step 8, making k equal to k +1 to update the clustering number, and jumping to step 6;
specifically, the number of clusters is increased.
Step 9, outputting a result of K-means cluster analysis based on the cluster number K, and obtaining the number of user meters in the distribution area meter box consistent with the cluster number K according to the determined cluster number K;
specifically, when the clustering termination condition is met, the result of the K-means clustering analysis based on the clustering number K can be output.
And step 10, determining a distribution area meter box to which any user meter belongs by combining the user meter ID addresses in the user meter working parameter samples in the clustering classification results, and obtaining the topological relation between the distribution area meter box and the user meters.
Further, during specific implementation, the meter box monitoring terminal modulates the address of each user meter into the working current, and when the number of the user meters in the meter box of the transformer area is obtained, the corresponding address of the user meter in the meter box of the transformer area is obtained at the same time.
In specific implementation, when the K-means cluster classification algorithm based on the CPSO is executed, after the collected three-phase working current, the address of the user table is modulated into the sampled three-phase working current, and when the concentrator obtains the number of the user tables in the table area meter box according to the method, the address corresponding to the user table in each table area meter box can be specifically obtained according to the address modulation mode. The address of the user table can be modulated into the working current by adopting the common technical means in the technical field, and the concentrator can specifically obtain the corresponding address of the user table according to the modulation method.
In the embodiment of the present invention, as can be seen from the above description, the result of the K-means cluster analysis includes the number of user table classifications, the cluster center, and the number of user tables in each cluster group. The corresponding relation between the user meter and the meter box in the transformer area can be determined through the ID address of the meter box monitoring terminal corresponding to the meter box monitoring terminal. One meter box monitoring terminal only corresponds to one platform district meter box, and in the process of communication between the meter box monitoring terminal and the concentrator, the concentrator can automatically read the ID address of the user meter, so that the matching relationship between the user meter and the platform district meter box to which the user meter belongs can be determined, and finally the topological relationship between the platform district meter box and the user meter is obtained.
Furthermore, a risk early warning model based on a deep convolutional neural network Alexnet is configured in a transformer outlet side monitoring terminal, a branch box monitoring terminal and/or a meter box monitoring terminal,
for the transformer outgoing line side monitoring terminal, processing the terminal temperature of the transformer outgoing line side terminal, the three-phase current value of the transformer outgoing line side and the three-phase leakage current of the transformer outgoing line side through a risk early warning model to determine a risk early warning value of the transformer outgoing line side;
processing the branch box wiring terminal temperature, the branch box three-phase current value and the branch box three-phase leakage current through a risk early warning model configured in the branch box monitoring terminal so as to determine a risk early warning value of the branch box;
and (4) a meter box monitoring terminal is monitored, and the temperature of a wiring terminal of the meter box in the area, the three-phase current value of the meter box in the area and the three-phase leakage current of the meter box in the area are processed through a risk early warning model configured in the meter box monitoring terminal so as to determine the risk early warning value of the meter box in the area.
In the embodiment of the invention, the risk early warning model based on the Alexnet can be configured in the transformer outlet side monitoring terminal, the branch box monitoring terminal and/or the meter box monitoring terminal, and can be specifically selected according to the requirement.
When a risk early warning model based on a deep convolutional neural network Alexenet is configured in the transformer outgoing line side monitoring terminal, the transformer outgoing line side monitoring terminal is subjected to treatment on the temperature of a terminal of the transformer outgoing line side, the three-phase current value of the transformer outgoing line side and the three-phase leakage current of the transformer outgoing line side through the risk early warning model, so that the risk early warning value of the transformer outgoing line side is determined.
When a risk early warning model based on a deep convolutional neural network Alexnet is configured in the branch box monitoring terminal, processing the branch box terminal temperature, the branch box three-phase current value and the branch box three-phase leakage current through the risk early warning model configured in the branch box monitoring terminal so as to determine the risk early warning value of the branch box;
when a risk early warning model based on the deep convolutional neural network Alexnet is configured in the meter box monitoring terminal, the meter box monitoring terminal is subjected to treatment on the temperature of a wiring terminal of a table area meter box, the three-phase current value of the table area meter box and the three-phase leakage current of the table area meter box through the risk early warning model configured in the meter box monitoring terminal, so that the risk early warning value of the table area meter box is determined.
In specific implementation, the specific conditions of temperature, three-phase current value and three-phase leakage current are all the same as those in the prior art, and are well known to those skilled in the art, and are not described herein again.
Further, when constructing the risk early warning model, the method comprises the following steps:
step S1, providing a neural network basic model based on a deep convolutional neural network Alexnet, and configuring model basic state parameters of the neural network basic model, wherein the configured model basic state parameters comprise precision control parameters and a learning rate;
specifically, the basic neural network model based on the deep convolutional neural network Alexnet is the existing commonly used Alexnet neural network, and the specific situation is well known to those skilled in the art. In order to obtain a required risk early warning model, basic state parameters of the model need to be configured.
In specific implementation, the precision control parameter can be 0.6-0.9, and generally, the precision control parameter can be about 0.8; the learning rate is 0.5 to 0.7.
S2, making a sample data set for training a basic model of the neural network, wherein the sample data set comprises working state parameters under various working states, the working states comprise a normal working state, a full load working state and a fault state, the working state parameters comprise working process parameters and early warning state parameters corresponding to the working process parameters, and the working process parameters comprise temperature, three-phase current and three-phase leakage current;
dividing the manufactured sample data set into a required training sample set and a required verification sample set, wherein the training sample set is used for training the basic neural network model, and the verification sample set is used for verifying the trained basic neural network model;
in specific implementation, after basic state parameters of a model are configured, a sample data set needs to be manufactured, for the sample data set, working state parameters under various working states are needed, the working states comprise a normal working state, a full-load working state and a fault state, the working state parameters comprise temperature, three-phase current and three-phase leakage current, and the sample data set needs to comprise the working state parameters under the normal working state, the working state parameters under the full-load working state and the working state parameters under the fault state.
The specific conditions of the normal operating state, the full load operating state and the fault state are based on the corresponding actual operating conditions, which are well known to those skilled in the art and will not be described herein again. Therefore, the risk early warning models based on the deep convolutional neural network Alexnet are configured in the transformer outgoing line side monitoring terminal, the branch box monitoring terminal and the meter box monitoring terminal, and corresponding sample data sets need to be respectively manufactured so as to meet the requirement that the transformer outgoing line side monitoring terminal, the branch box monitoring terminal and the meter box monitoring terminal carry out required risk monitoring.
In the embodiment of the invention, the sample data set comprises a plurality of sample data, in order to obtain the sample data, sampling basic data corresponding to the sample data needs to be collected firstly, and the sampling basic data is processed to obtain the sample data, wherein the processing of the sampling basic data comprises the steps of characteristic value extraction, oscillation mode screening and preprocessing which are sequentially carried out.
For any sampling basic data, characteristic value extraction is needed, and the characteristic value is a value representing the physical characteristics of the sampling basic data. In specific implementation, for three-phase current, the characteristic values are effective values and peak values of current output of the equipment in a full-load state, and the current value of the household meter in a long-time limit working state can be represented. For the characteristic value of the temperature, a product manual of a monitoring user table on which the operating maximum temperature or the threshold temperature of the terminal is located may be generally referred to take the operating maximum temperature or the threshold temperature as the characteristic value of the temperature. For the three-phase leakage current, the effective value or the peak value of the three-phase leakage current can be inquired into a user table and a product manual of a monitoring terminal, and the characteristic value of the leakage current can be determined by referring to the relevant technical specifications of the low-voltage distribution network. Therefore, according to the characteristics corresponding to the temperature, the three-phase current and the three-phase leakage current, the characteristic value of the sampling basic data can be extracted, and after the characteristic extraction is carried out, the sampling characteristic value data is obtained. Specifically, the temperature specifically refers to an inlet end temperature and a corresponding outlet end temperature.
After the sampling characteristic value data is obtained, carrying out oscillation mode screening on the sampling characteristic value data, wherein the oscillation mode screening is equivalent to adding a white noise signal on the basis of characteristic value extraction, and processing after adding the white noise signal to generate signal data containing various oscillation modes; according to the sensitivity of the electric power equipment to which the low-voltage distribution area belongs, a required oscillation damping ratio is selected, for example, the range of the oscillation damping ratio is [0, 1], the processing after white noise is added can be specifically realized by selecting a table of sampling characteristic values and selecting a signal with the maximum standard deviation as a signal after the oscillation mode screening. In specific implementation, when the sampling characteristic value data is subjected to oscillation mode screening, the specific oscillation mode screening manner is consistent with that of the prior art, and is specifically known to those skilled in the art, and is not described herein again.
And after the oscillation mode is screened, carrying out a preprocessing step, wherein the preprocessing step specifically comprises normalization processing, namely carrying out corresponding normalization processing on the data after the oscillation mode is screened, and obtaining sample data after normalization, wherein the value ranges of the temperature, the three-phase current and the three-phase leakage current in the sample data set are all [0, 1 ].
In specific implementation, for the early warning state parameter in the working state parameters, the same processing procedure for the working process parameters needs to be performed, that is, the steps of feature value extraction, oscillation mode screening, and preprocessing are performed. For the early warning state parameters, the incoming line temperature abnormal state parameters, the outgoing line temperature abnormal state parameters, the three-phase current abnormal state parameters, and the three-phase leakage current abnormal state parameters may be specifically used, wherein the abnormal state is related to an actual working scene, and the like, and is specifically related to a case where the abnormality is determined as needed, for example, the incoming line temperature abnormality may be determined as abnormal when the temperature is higher than a set temperature, and the specific cases of the three-phase current abnormality and the three-phase leakage current abnormality may refer to the description of the incoming line temperature abnormality, which is specifically known to those skilled in the art, and will not be described herein again.
The created sample data set is divided into a training sample set and a verification sample set as required, and the specific situations of the training sample set and the verification sample set are consistent with those in the prior art, which are well known to those skilled in the art and will not be described herein again.
Step S3, training the neural network basic model by using a training sample set, and verifying the trained neural network basic model by using a verification sample set when a training termination condition is reached;
and when the verification of the neural network basic model by using the verification sample set is matched with the verification conditions, obtaining a risk early warning model based on the deep convolution neural network A1exnet by using the trained neural network basic model.
During specific implementation, training termination conditions are set according to risk early warning requirements, and the specific conditions of the training termination conditions can be selected according to needs. And verifying the trained neural network basic model by using a verification sample set so as to evaluate the trained neural network basic model, wherein the existing common standard can be adopted during evaluation, and the trained neural network basic model can be configured into a risk early warning model based on the deep convolution neural network Alexnet after meeting the evaluation standard.
Training a basic neural network model through working process parameters and early warning state parameters in a sample data set, and obtaining a risk early warning model based on the deep convolutional neural network Alexnet, so that a risk early warning state value consistent with the early warning state parameters in the sample data set can be obtained during early warning. In specific implementation, the risk early warning state value is generally a data matrix, each early warning mode in the risk early warning state value is independent of each other, and each row of the data matrix corresponds to one fault mode. The number of rows of the data matrix corresponding to the risk early warning state value can be selected according to actual needs, so that the number of the early warning fault modes contained in the early warning state parameters in the sample data set can be correspondingly consistent, namely, the required risk early warning state value can be output after training according to the actual early warning condition.
In specific implementation, when the risk early warning is performed by using the risk early warning model, the risk early warning value of the transformer outgoing line side is the risk early warning state value of the transformer outgoing line side for the transformer outgoing line side monitoring terminal. Of course, corresponding to the branch box monitoring terminal and the meter box monitoring terminal, the risk early warning values are corresponding risk early warning state values, and the specific situation may refer to the above description, which is not repeated herein.
Further, according to kirchhoff's current law, the sum of the currents flowing into the nodes is equal to the sum of the currents flowing out of the nodes, and the voltage of the nodes is kept constant. For one land, current will flow from the land node to the next branch node, which flows step by step along the physical line, to the user charge side. Setting the main node current and power of the zone concentrator as (IO, PO), and the current and power flowing out of each node as (I)A1，PA1)，(IA2，PA2)，...，(IAn，PAn) Then, there are:
I0＝IA1+IA2+...+IAn+ΔIA
P0＝PA1+PA2+...+PAn+ΔPA
explaining a two-stage branch outlet end node and a meter box stage inlet end node in the nodes, according to a station area topology identification result, the power and the current among the nodes at all stages should meet the following charge relation:
(IA11，PA11)＝(IA111+IA112+IA113+IA114+ΔIA11，
PA111+PA112+PA113+PA114+ΔPA11)
wherein, Delta IA11、ΔPA11The current and power loss between the secondary branch outgoing line end node A11 and the meter box level incoming line end node line are respectively.
In the embodiment of the invention, the concentrator can calculate the power loss of each stage of line in the transformer area and the total power loss of the transformer area according to the principle, and calculate the line loss rate according to the power loss value.
Further, still include the environmental monitoring device who is connected with table case monitor terminal adaptation electricity, utilize environmental monitoring device can monitor the environmental state parameter of table case monitor terminal place environment to transmit the environmental state parameter who monitors to the concentrator.
When specifically implementing, table case monitor terminal still possesses the ability of thing networking interface, can be connected environment detection device and table case monitor terminal to can directly acquire the environmental status parameter monitoring of table case monitor terminal place environment, link status parameter can be link states such as temperature, humidity, specifically can select as required. After the ring section state parameters are transmitted to the concentrator, the environment state of the environment where each meter box monitoring terminal is located can be determined by the concentration area.
In conclusion, the method for monitoring the power distribution state of the low-voltage transformer area can be obtained, and the concentrator and the transformer area power distribution monitoring device which is communicated with the concentrator on the basis of the HPLC network are provided; the distribution monitoring device comprises a transformer outgoing side monitoring terminal, a plurality of branch box monitoring terminals and a plurality of meter box monitoring terminals, wherein the transformer outgoing side monitoring terminals are in adaptive connection with transformer outgoing terminals of the distribution area;
after the concentrator communicates with a transformer outgoing line side monitoring terminal, a branch box monitoring terminal and a meter box monitoring terminal on the basis of an HPLC network, the concentrator determines a distribution topology state of a transformer area where the concentrator is located, the determined distribution topology state comprises a transformer area household variable relation, a transformer area branch topology and a transformer area meter box topology, and when the concentrator determines the transformer area meter box topology, the number of user meters in a transformer area meter box corresponding to any meter box monitoring terminal is determined by adopting a CPSO-based K-means cluster classification algorithm.
Specifically, the concentrator, the transformer outgoing line side monitoring terminal in the power distribution detection device, the branch box monitoring terminal, and the meter box monitoring terminal are specifically matched with each other, so that the above description can be referred to the method for realizing the state monitoring, and the details are not repeated here.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210477862.8A CN114696466B (en) | 2022-04-29 | 2022-04-29 | State monitoring system and method suitable for low voltage distribution area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210477862.8A CN114696466B (en) | 2022-04-29 | 2022-04-29 | State monitoring system and method suitable for low voltage distribution area |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114696466A true CN114696466A (en) | 2022-07-01 |
| CN114696466B CN114696466B (en) | 2024-07-19 |
Family
ID=82145571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210477862.8A Active CN114696466B (en) | 2022-04-29 | 2022-04-29 | State monitoring system and method suitable for low voltage distribution area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114696466B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115118015A (en) * | 2022-07-22 | 2022-09-27 | 安徽明生恒卓科技有限公司 | Platform district power supply stability monitoring system based on fuse terminal |
| CN115296421A (en) * | 2022-09-28 | 2022-11-04 | 南方电网数字电网研究院有限公司 | Power distribution system monitoring method, device, equipment and medium based on carbon metering chip |
| CN115347918A (en) * | 2022-08-15 | 2022-11-15 | 北京航天鸣镝科技有限公司 | Low-voltage transformer area meter box physical topology identification method and system |
| CN115423046A (en) * | 2022-10-25 | 2022-12-02 | 安徽南瑞中天电力电子有限公司 | Terminal low-voltage distribution area topology identification method and device based on clustering algorithm |
| CN116526680A (en) * | 2023-05-31 | 2023-08-01 | 威海双联电气有限公司 | Control method and device for high-low voltage power distribution cabinet cluster |
| CN116846085A (en) * | 2023-09-04 | 2023-10-03 | 山西绿柳科技有限公司 | A self-sensing monitoring intelligent distribution box control system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110943450A (en) * | 2019-12-12 | 2020-03-31 | 山东电工电气集团有限公司 | Platform area automatic topology line loss analysis method based on Internet of things |
| CN112510707A (en) * | 2020-12-09 | 2021-03-16 | 国网浙江省电力有限公司电力科学研究院 | Power distribution station power impedance topological graph generation method and system |
| CN113572164A (en) * | 2021-08-06 | 2021-10-29 | 国网四川省电力公司营销服务中心 | Power distribution network region identification method based on k-means cluster analysis |
| WO2022016903A1 (en) * | 2020-07-24 | 2022-01-27 | 石家庄科林电气股份有限公司 | Automatic low-voltage area topology identification method |
| CN114123486A (en) * | 2021-10-25 | 2022-03-01 | 国网浙江省电力有限公司诸暨市供电公司 | Intelligent low-voltage distribution area system with whole network sensing function |
-
2022
- 2022-04-29 CN CN202210477862.8A patent/CN114696466B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110943450A (en) * | 2019-12-12 | 2020-03-31 | 山东电工电气集团有限公司 | Platform area automatic topology line loss analysis method based on Internet of things |
| WO2022016903A1 (en) * | 2020-07-24 | 2022-01-27 | 石家庄科林电气股份有限公司 | Automatic low-voltage area topology identification method |
| CN112510707A (en) * | 2020-12-09 | 2021-03-16 | 国网浙江省电力有限公司电力科学研究院 | Power distribution station power impedance topological graph generation method and system |
| CN113572164A (en) * | 2021-08-06 | 2021-10-29 | 国网四川省电力公司营销服务中心 | Power distribution network region identification method based on k-means cluster analysis |
| CN114123486A (en) * | 2021-10-25 | 2022-03-01 | 国网浙江省电力有限公司诸暨市供电公司 | Intelligent low-voltage distribution area system with whole network sensing function |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115118015A (en) * | 2022-07-22 | 2022-09-27 | 安徽明生恒卓科技有限公司 | Platform district power supply stability monitoring system based on fuse terminal |
| CN115118015B (en) * | 2022-07-22 | 2024-12-06 | 安徽明生恒卓科技有限公司 | A power supply stability monitoring system based on fusion terminal |
| CN115347918A (en) * | 2022-08-15 | 2022-11-15 | 北京航天鸣镝科技有限公司 | Low-voltage transformer area meter box physical topology identification method and system |
| CN115347918B (en) * | 2022-08-15 | 2025-12-26 | 北京航天鸣镝科技有限公司 | Physical Topology Identification Method and System for Low-Voltage Distribution Area Meter Boxes |
| CN115296421A (en) * | 2022-09-28 | 2022-11-04 | 南方电网数字电网研究院有限公司 | Power distribution system monitoring method, device, equipment and medium based on carbon metering chip |
| CN115423046A (en) * | 2022-10-25 | 2022-12-02 | 安徽南瑞中天电力电子有限公司 | Terminal low-voltage distribution area topology identification method and device based on clustering algorithm |
| CN116526680A (en) * | 2023-05-31 | 2023-08-01 | 威海双联电气有限公司 | Control method and device for high-low voltage power distribution cabinet cluster |
| CN116526680B (en) * | 2023-05-31 | 2023-09-19 | 威海双联电气有限公司 | Control method and device for high-low voltage power distribution cabinet cluster |
| CN116846085A (en) * | 2023-09-04 | 2023-10-03 | 山西绿柳科技有限公司 | A self-sensing monitoring intelligent distribution box control system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114696466B (en) | 2024-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114696466A (en) | State monitoring system and method suitable for low-voltage transformer area power distribution | |
| Damousis et al. | A fuzzy expert system for the forecasting of wind speed and power generation in wind farms | |
| Damousis et al. | A fuzzy model for wind speed prediction and power generation in wind parks using spatial correlation | |
| Liu et al. | Wind power plant prediction by using neural networks | |
| CN106569030B (en) | Alarm threshold optimization method and device in a kind of electrical energy measurement abnormity diagnosis | |
| CN107609783A (en) | The method and system that a kind of intelligent electric energy meter combination property based on data mining is assessed | |
| CN112132488B (en) | A coupled modeling-optimization-optimization reservoir ecological dispatch method | |
| CN109301832B (en) | Section flow optimization control method based on N-1 static safety constraint | |
| CN118017509A (en) | Parallel optimization method for large-scale distribution networks based on digital twin space | |
| CN118611117A (en) | A collaborative planning method for active distribution networks based on scenario fuzzy clustering and multi-agent evolutionary game | |
| CN112241812B (en) | Low-voltage distribution network topology identification method based on unilateral optimization and genetic algorithm collaboration | |
| CN110705887A (en) | A comprehensive evaluation method of low-voltage station operating state based on neural network model | |
| Venkatesh et al. | Multi-objective evolutionary programming for economic emission dispatch problem | |
| CN118801355A (en) | Power dispatch optimization method and system based on digital twin | |
| CN116502922B (en) | Power grid stability analysis system based on group intelligent algorithm | |
| CN116203351B (en) | Method and system for detecting abnormal line impedance | |
| Jeyakumar et al. | Application of multi objective evolutionary programming to combined economic emission dispatch problem | |
| Mohammadi et al. | Multi-objective optimization of decision trees for power system voltage security assessment | |
| CN120812800A (en) | Intelligent control method and system for solar street lamp | |
| CN115313361A (en) | Joint optimization control method and device for large-scale adjustable resources | |
| CN112686404B (en) | Power distribution network fault first-aid repair-based collaborative optimization method | |
| CN114089118A (en) | Fault location method for smart substation based on gated cyclic unit network | |
| Chen et al. | A method for distribution network topology identification based on mutual information and DNN | |
| CN118868061A (en) | A regional inertia online monitoring method and system based on convolutional neural network | |
| CN111898309B (en) | Photovoltaic intelligent edge terminal optimized layout method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |