CN120110008A

CN120110008A - A three-dimensional digital intelligent monitoring system suitable for photovoltaic power stations

Info

Publication number: CN120110008A
Application number: CN202510260528.0A
Authority: CN
Inventors: 张文敏; 孙鹏飞; 朱德磊
Original assignee: Zhongke Yunshang Nanjing Intelligent Technology Co ltd; Beijing Zhongke Lifeng Technology Co ltd
Current assignee: Zhongke Yunshang Nanjing Intelligent Technology Co ltd; Beijing Zhongke Lifeng Technology Co ltd
Priority date: 2025-03-06
Filing date: 2025-03-06
Publication date: 2025-06-06

Abstract

The invention discloses a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station, which relates to the technical field of photovoltaic power station monitoring and comprises a model construction unit, a hanging storage unit, a three-dimensional overview unit, an abnormality identification unit and a monitoring alarm unit, wherein the model construction unit is used for acquiring multi-dimensional electrical equipment attribute parameters and constructing an equipment model, the hanging storage unit is used for hanging the equipment model to real-time operation data and carrying out distributed storage on the real-time operation data, the three-dimensional overview unit is used for constructing a three-dimensional model of the photovoltaic power station and dynamically displaying the real-time operation state of each equipment model of the photovoltaic power station, the abnormality identification unit is used for constructing an abnormality identification model based on deep learning and analyzing potential abnormality existing in the photovoltaic power station, and the monitoring alarm unit is used for monitoring the operation parameters of the photovoltaic power station in real time and realizing alarm monitoring. According to the invention, through three-dimensional digital twin modeling and real-time data hooking, a comprehensive power station information display platform is constructed, dynamic monitoring of power station equipment can be realized, and the efficiency and the accuracy of monitoring the photovoltaic power station are improved.

Description

Three-dimensional digital intelligent monitoring system suitable for photovoltaic power station

Technical Field

The invention relates to the technical field of photovoltaic power station monitoring, in particular to a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station.

Background

In photovoltaic power plant field, realize the comprehensive, accurate control to power plant equipment have the vital meaning to guarantee power plant steady operation, improve generating efficiency and extension equipment life. With the continuous expansion of the scale of photovoltaic power generation and the increasing maturity of technology, the variety and the quantity of equipment in a power station are rapidly increased, and the problems of incomplete information acquisition and imprecise monitoring are exposed by a traditional monitoring system, so that the requirements of a modern power station on data instantaneity and accuracy are difficult to meet.

First, conventional monitoring systems do not collect information comprehensively. Most systems rely on only a small number of low-precision sensors, and are difficult to cover all areas of a power station and key equipment, such as photovoltaic modules, inverters, junction boxes and the like, so that data of the running states of part of the equipment are missing or inaccurate. In addition, the data interfaces and communication protocols between the devices are administrative and lack of unified standards, further impairing the integrity and consistency of data collection.

Second, the lack of real-time monitoring capability is also a significant shortboard of the current technology. The existing monitoring system has the common delay problem in the data transmission process, has low updating frequency, and cannot intuitively and real-timely reflect the overall operation state of the power station and the fine change of each device. The data hysteresis phenomenon causes the judgment and identification of the abnormal condition of the equipment to lack timeliness, thereby influencing the judgment basis of the whole safe operation of the power station.

In addition, data integration and analysis capabilities are limited. The traditional system can only realize basic data record and simple statistics, and lacks the functions of deep mining and trend analysis on mass data. Because the data formats of all the devices are not uniform, and the data storage is dependent on local devices, the data integration efficiency is low, and the historical data inquiry and the cross-device data comparison are difficult. Meanwhile, the system lacks a perfect early warning mechanism when facing abnormal data, and early identification and risk prejudgment of hidden danger are difficult to realize.

In general, the comprehensive and accurate monitoring system based on the advanced information technology is constructed, so that the defects of the traditional photovoltaic power station monitoring system in the aspects of information acquisition and data analysis are overcome, and solid guarantee is provided for safe and efficient operation of the power station.

Disclosure of Invention

Based on the above, it is necessary to provide a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station.

The invention provides a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station, which comprises the following components:

the model building unit is used for obtaining multi-dimensional electrical equipment attribute parameters and building an equipment model;

The hooking storage unit is used for hooking the equipment model to the real-time operation data and carrying out distributed storage on the real-time operation data;

the three-dimensional overview unit is used for constructing a three-dimensional model of the photovoltaic power station and dynamically displaying the real-time running state of each equipment model of the photovoltaic power station;

the anomaly identification unit is used for constructing an anomaly identification model based on deep learning and analyzing potential anomalies of the photovoltaic power station;

and the monitoring alarm unit is used for monitoring the operation parameters of the photovoltaic power station in real time and realizing alarm monitoring.

Further, the equipment models comprise a power station model, a booster station model, a box transformer model, an inverter model, a combiner box group string model, an optical resource model and other equipment models.

Further, the hooking storage unit includes:

The dynamic hooking module is used for creating a data interface, communicating data interaction between each equipment model and real-time operation data, constructing a hierarchical communication architecture, and monitoring data hooking and updating in real time;

and the storage management module is used for dispersedly storing real-time operation data, meteorological data and model data of the photovoltaic power station to a plurality of nodes by adopting a distributed storage technology, introducing a redundancy backup mechanism and an erasure code technology, and maintaining long-term availability and integrity of the data.

Further, the three-dimensional overview unit includes:

the three-dimensional modeling module is used for combining oblique photography and vector modeling to construct a three-dimensional model of the photovoltaic power station and supporting dynamic display of the real-time data driving model;

the road network space module is used for digitally reconstructing the road network and the electrical equipment of the factory area based on the unmanned aerial vehicle aerial image data to form an electronic map space database;

And the map overview module is used for constructing a three-dimensional digital scene map and displaying the real-time running states of various equipment models in the photoelectric power station.

Further, the abnormality identifying unit includes:

The model construction training module is used for collecting historical operation data for preprocessing, constructing an anomaly identification model of the long-term and short-term memory network model based on deep learning and carrying out model training;

The model early warning analysis module is used for setting an input variable and an output variable, and predicting the running state of the equipment model at the future moment by using the trained abnormal recognition model;

The data analysis and quantization module is used for carrying out differential processing on the time series data, capturing trend characteristics of the operation data by utilizing an autoregressive and moving average model and quantifying the performance degradation degree of the equipment;

the power trend prediction module is used for acquiring historical photovoltaic data of the photovoltaic module, predicting the power attenuation trend of the photovoltaic module in a future time period by using the anomaly identification model, drawing a power attenuation trend curve, and responding to equipment maintenance measures according to a trend prediction result.

Further, performing differential processing on the time series data, capturing trend characteristics of the operation data by utilizing an autoregressive and moving average model, and quantifying the performance degradation degree of the equipment comprises:

converting the real-time operation data of the acquired equipment model into time sequence data, carrying out stability test on the time sequence data, and executing differential processing if the stability threshold is not met;

Setting model parameters of a summation autoregressive moving average model, determining optimal model parameters through a minimum red pool information criterion, fitting time series data by using the optimized summation autoregressive moving average model to obtain a model predicted value, and calculating an error between the model predicted value and a true value;

and analyzing trend items and period items of the time series data based on fitting results, and quantifying the performance degradation degree of the equipment model corresponding to the electrical equipment according to the change rate of the trend items.

Further, the monitoring alarm unit includes:

The state monitoring module is used for establishing a hierarchical relation model of each equipment model in the photovoltaic power station, performing state monitoring and state adjustment of the electrical equipment based on the hierarchical relation, and performing collaborative optimization with the anomaly identification unit to generate a visual chart of state monitoring data;

And the intelligent alarm module is used for acquiring monitoring data of each electrical device in real time by utilizing the sensor, judging whether the device model triggers an alarm response or not, and analyzing the fault reason.

Further, the status monitoring module includes:

the hierarchical relation sub-module is used for establishing a hierarchical relation model for sequentially connecting the string, the combiner box, the inverter, the box transformer and the booster station;

The monitoring and adjusting sub-module is used for executing the state monitoring of the electrical equipment based on the upper-level and lower-level relation and adjusting the operation strategy of the upper-level equipment according to the state of the lower-level equipment;

The sharing coordination sub-module is used for synchronizing real-time operation data and monitoring data of the electrical equipment of each level to the anomaly identification unit, acquiring a trend prediction result output by the anomaly identification model, and executing response control between upper and lower levels of equipment according to faults or anomalies of the electrical equipment;

and the chart analysis sub-module is used for drawing a visual chart according to the operation data of the electrical equipment.

Further, performing electrical device status monitoring based on the upper-lower relationship, and adjusting the upper-level device operation policy according to the lower-level device status includes:

acquiring real-time operation data of each electrical device obtained through real-time monitoring, and judging that the corresponding electrical device has a fault phenomenon if any one item of data in the real-time operation data does not meet a safety threshold value;

If the group string current uploaded by the junction box has fault information, the junction box is called to reduce the input power of a corresponding channel, and if the plurality of junction boxes have fault information, the output power of the inverter is reduced and alarm information is sent;

If the box transformer detects that the output of the inverter has fault information, the tap position of the box transformer is adjusted, the transformation ratio is adjusted, and if the fault information is larger than a preset threshold value, the box transformer is disconnected with the inverter;

and if the box-type fault information exceeds the quantity threshold, the output power of the photovoltaic power station is reduced, and the power supply of a fault area is cut off.

Further, the intelligent alarm module includes:

The early warning analysis sub-module is used for collecting monitoring data of each electrical device in real time, judging and analyzing whether each electrical device has a fault or not through threshold judgment, and triggering early warning and reminding after the fault is found;

The threshold optimization sub-module is used for analyzing the change rules of the operation of the electrical equipment under different working conditions and environments through the anomaly identification model and dynamically adjusting and judging the threshold;

the early warning support sub-module is used for acquiring a trend prediction result of the anomaly identification model and fault monitoring information of the electrical equipment, and triggering early warning reminding and maintenance response in advance;

And the analysis assisting sub-module is used for matching fault reasons and outputting a fault maintenance direction.

The beneficial effects of the invention are as follows:

1. The invention aims to solve the problems of incomplete information and unsophisticated monitoring of the traditional photovoltaic power station monitoring system, builds a comprehensive power station information display platform through three-dimensional digital twin modeling and real-time data hanging, can realize dynamic monitoring of power station equipment, effectively solves the problems, improves the efficiency and the accuracy of monitoring the photovoltaic power station, not only makes up the defects of the traditional photovoltaic power station monitoring system in information acquisition and data analysis, but also provides solid guarantee for safe and efficient operation of the power station.

2. The invention realizes the omnibearing digital monitoring of the photovoltaic power station, acquires the attribute of the multidimensional electrical equipment and builds an equipment model through the model building unit, then connects the model with real-time operation data, efficiently manages the data by using a distributed storage technology, ensures the timeliness and accuracy of information transmission, realizes the dynamic and visual display of the states of all the equipment by the three-dimensional overview unit, enables operation staff to quickly know the overall operation condition of the power station, captures potential faults in advance by the abnormal recognition unit based on deep learning, and can send early warning in the first time in cooperation with the real-time alarm monitoring unit, thereby reducing the risk of accidents.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation on the invention. In the drawings:

Fig. 1 is a system block diagram of a three-dimensional digital intelligent monitoring system suitable for use in a photovoltaic power plant in accordance with an embodiment of the present invention.

The reference numeral 1, a model construction unit, 2, a hanging storage unit, 3, a three-dimensional overview unit, 4, an anomaly identification unit and 5, a monitoring alarm unit.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

Referring to fig. 1, a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station is provided, including:

The model building unit 1 is used for obtaining the multi-dimensional electrical equipment attribute parameters and building an equipment model.

In the description of the present invention, the equipment model includes a power station model, a booster station model, a box transformer model, an inverter model, a combiner box string model, an optical resource model, and other equipment models.

Specifically, the device model and the corresponding electrical device attribute parameters thereof comprise the following aspects:

1. and the power station model comprises electric energy metering information and gateway meter data, and realizes acquisition and processing of electric energy metering data and meter state information.

2. The booster station model comprises remote pulse data, remote control data and remote measurement signals, and covers breaker switching instructions of main equipment and a main loop.

3. The box transformer model comprises the electric quantity, switching value and transformer body signals of box transformer equipment, such as all telemetry of box transformer, namely three-phase current, three-phase voltage and power.

4. The inverter model comprises the attributes of the electric quantity and the switching value of the AC/DC side of the inverter and the signals of the inverter body, such as input voltage, input current, output active power and the like.

5. The combiner box/cluster model attributes include line voltage, phase current, power factor, active power, reactive power, etc.

6. And the light resource model comprises the attributes including solar irradiance, solar irradiation quantity, wind speed, wind direction, temperature and power generation power prediction result data.

7. The equipment model comprises other data, such as electric energy data, such as generated energy, electric consumption and the like, equipment model configuration data, information related to parameter setting, model specification, installation position and the like of equipment, and operation and maintenance related data such as maintenance record, fault history and the like of the equipment, wherein the data can provide comprehensive data support for the operation state evaluation, fault diagnosis and maintenance decision of the equipment.

In addition, the detailed steps of the construction of the various equipment models comprise the following aspects:

1. the detailed steps of the power station model are as follows:

And (3) corresponding data acquisition equipment is deployed, communication connection is established between the data acquisition equipment and the electric energy metering device and the gateway meter, and electric energy metering information and gateway meter data are obtained.

And preprocessing the acquired data by adopting a specific data processing algorithm, such as removing abnormal values, smoothing the data and the like, so as to ensure the accuracy and stability of the data.

The processed data is stored in a database for subsequent analysis and presentation.

2. The booster station model comprises the following detailed steps:

and collecting remote pulse data, remote control data and remote measurement signals by using the sensor and the communication module.

Analyzing and verifying the collected breaker switching instruction, and ensuring the accuracy and the completeness of the instruction.

And integrating the related data into the booster station model, and updating the model state according to the real-time data.

3. The detailed steps of the box transformer model are as follows:

Various sensors are installed on the box transformer equipment, and the electric quantity such as three-phase current, three-phase voltage, power and the like, the switching value and the transformer body signal are monitored in real time.

The collected data is transmitted to a data processing center through a communication network.

And analyzing and processing the data, judging the running state of the box-type substation equipment, and feeding the result back to the box-type substation model.

4. Inverter model detailed steps:

And the measuring equipment is connected with the AC/DC side of the inverter to acquire the electric quantity and switching value of input voltage, input current, output active power and the like and the signals of the inverter body.

And analyzing the collected data in real time, and monitoring the operation efficiency and stability of the inverter.

And correlating the analysis result with the inverter model to realize dynamic updating of the model.

5. The combiner box/string model detailed steps:

corresponding monitoring devices are arranged on the combiner box and the group string, and data such as line voltage, phase current, power factor, active power, reactive power and the like are collected.

And sorting and classifying the data, and establishing a corresponding relation between the data and the model.

And adjusting the model state in real time according to the data change, and intuitively displaying the running conditions of the combiner box and the group string.

6. Detailed steps of the optical resource model:

Meteorological data such as solar irradiance, wind speed, wind direction, temperature and the like are collected through equipment such as a weather station and the like.

And a power generation power prediction algorithm is used for predicting the power generation power by combining historical data and current meteorological conditions.

The collected and predicted data is integrated into the optical resource model to provide a reference for the operation of the power station.

7. Other data detail steps of the equipment model:

the power data of the power station is collected periodically, including power generation, power consumption, etc.

And updating and maintaining the equipment model configuration data according to the actual configuration condition of the power station equipment.

Integrating and storing the electric energy data and the equipment model configuration data, and providing support for subsequent analysis.

And the hooking storage unit 2 is used for hooking the equipment model to the real-time operation data and carrying out distributed storage on the real-time operation data.

In the description of the present invention, the hooking storage unit 2 includes a dynamic hooking module (not shown) and a storage management module (not shown).

The dynamic hooking module is used for creating a data interface, communicating data interaction between each equipment model and real-time operation data, constructing a hierarchical communication architecture, monitoring data hooking and updating in real time, and finally collecting production operation and maintenance derived data in the station.

Specifically, the function of the dynamic hooking module is to establish a data interface to ensure that each equipment model and the real-time operation data system can realize data interaction, and the content and principle of the module comprise the following contents:

1. data interface design and communication mode:

1.1, interface type:

and the standardized protocol is OPC UA (for industrial equipment communication), MQTT (lightweight Internet of things protocol) or Modbus TCP (power equipment common protocol), and is adapted to the communication requirements of different equipment models.

API interaction, namely realizing cross-platform data interaction through RESTful API aiming at cloud data synchronization, and supporting JSON or Protobuf formats.

2. Connection relation between equipment models:

2.1, hierarchical communication architecture:

bottom layer equipment (combiner box/string model) →inverter model (string current, voltage are collected via RS-485 or CAN bus).

Inverter model→box-section model (ac side power data is transmitted via ethernet).

Tank transformer model→booster station model (transmitting breaker status, transformer temperature via IEC 61850 protocol).

Optical resource model→plant model (pushing weather forecast data to central database via HTTP).

2.2, Bidirectional interaction, wherein the booster station model can send a remote control instruction to the box transformer model (such as opening and closing of a breaker), and the inverter model receives a power scheduling instruction of the power station model.

3. Data hooking rules and update mechanism:

3.1, mapping rule:

And the unique device identifier is that a globally unique ID (such as UUID) is allocated to each device model, so that accurate binding of data and physical devices is realized.

Data tagging, namely supporting semantic query and association analysis by adopting a unified naming rule (such as/photovoltaic area/inverter 001/alternating voltage).

3.2, Updating strategies:

Real-time classification:

millisecond level: breaker status, fault signal (alarm triggered directly by edge computing device).

Second level, electrical quantity data (current, voltage), meteorological data (buffered by a time sequence database InfluxDB).

Minute-level, equipment health status, power generation efficiency analysis (updated by batch processing).

Incremental synchronization, namely transmitting only change data (Delta Update), and reducing network load.

And 3.3, transmission encryption, namely encrypting a communication link by using TLS1.3, wherein the equipment side adopts a Hardware Security Module (HSM) to store the secret key.

Sensitive instructions, such as remote control operations, require verification of source legitimacy by digital signature (ECDSA algorithm).

And 3.4, authority control, namely role-based access control (RBAC), wherein an operation and maintenance personnel can only view the data of the area, and a dispatcher has remote operation authority.

And (3) auditing operation logs, namely recording data modification and instruction issuing records, and supporting the tamper resistance of the blockchain storage card.

And (3) formulating a data hooking rule and a data hooking flow, and defining the corresponding relation and updating frequency of the data.

Dynamic data hanging (taking inverter model as an example)

4. And data acquisition, namely reading the electric quantity of the AC-DC side of the inverter through a Modbus TCP protocol, wherein the frequency is 1 second/time. And filtering abnormal data (such as transient fluctuation) by using an edge gateway, and uploading the smoothed abnormal data to the cloud.

And 4.1, binding the model, namely in a three-dimensional visual interface, associating the inverter entity model with a real-time data stream, and dynamically updating a power curve and an efficiency thermodynamic diagram.

And 4.2, exception handling, namely if the continuous 3 times of acquisition fails, automatically switching to a redundant communication channel (such as a 4G standby link). When data is lost, it is interpolated (linear or ARIMA predicted) based on historical data, temporarily filled, and labeled "estimated data".

4.3, Cross-model data fusion:

space-time alignment, namely adding uniform time stamps to meteorological data (an optical resource model) and power generation data (a power station model), and supporting correlation analysis (such as irradiance-power deviation detection).

And (3) event-driven updating, namely automatically triggering an unmanned aerial vehicle inspection task (a road network space model) when the light resource model predicts strong wind weather, and adjusting the insulation monitoring frequency of the string model. And (3) hanging the real-time operation data on a corresponding equipment model according to the rule, and monitoring and updating the data in real time.

Specifically, the invention adopts a distributed storage technology to store the operation data, the meteorological data, the model data and the like of the power station equipment in a plurality of nodes in a dispersed way. By using a redundant backup mechanism, such as multi-copy storage, data can still be read normally when part of nodes fail, and the risk of data loss is reduced by 99%. Meanwhile, by combining an erasure code technology, the storage efficiency and the data fault tolerance are further improved, and the long-term integrity and usability of the data are ensured.

For example, a distributed storage system in combination with a distributed computing framework (e.g., APACHE SPARK) may enable parallel processing of large-scale data. When the mass equipment operation data is processed to perform power generation trend analysis and fault diagnosis, the processing speed is improved by 5-10 times. By means of the distributed index technology, specific data, such as abnormal data of specific equipment in a certain time period, are rapidly located and retrieved, response time is shortened from a minute level to a second level, and system instantaneity and operation and maintenance efficiency are improved.

And the three-dimensional overview unit 3 is used for constructing a three-dimensional model of the photovoltaic power station and dynamically displaying the real-time running state of each equipment model of the photovoltaic power station.

In the description of the present invention, the three-dimensional overview unit 3 includes a three-dimensional modeling module (not shown in the drawings), a road network space module (not shown in the drawings), and a map overview module (not shown in the drawings).

And the three-dimensional modeling module is used for combining oblique photography and vector modeling to construct a three-dimensional model of the photovoltaic power station and supporting dynamic display of the real-time data driving model.

Specifically, the three-dimensional modeling module uses the unmanned aerial vehicle to perform oblique photography, and obtains image data of different angles of the photovoltaic power station. The vector modeling technology is adopted to model elements such as a photovoltaic module, a photovoltaic bracket, an inverter, a box transformer, a booster station, electrical equipment, facilities, a weather station (if any), an unmanned aerial vehicle hangar, related constructions, structures, a terrain environment and the like.

And fusing the oblique photographic data with the vector model to construct a fine photovoltaic power station three-dimensional model. And setting a real-time data interface to enable the model to be dynamically displayed according to the real-time data.

And the road network space module is used for digitally reconstructing the road network and the electrical equipment of the factory area based on the unmanned aerial vehicle aerial image data to form an electronic map space database.

Specifically, the image data of the factory road network and the electrical equipment are obtained through unmanned aerial vehicle aerial photography. And analyzing and processing the image data by using an image processing and pattern recognition technology, and extracting the characteristic information of the road network and the electrical equipment. And constructing an electronic map space database according to the extracted characteristic information, and realizing the digital reconstruction of the factory road network and the electrical equipment.

Specifically, the map overview module constructs a three-dimensional digital scene map to display real-time running states of various equipment models in the photoelectric power station, and the method comprises the following steps:

1. Unmanned aerial vehicle modeling detailed processing flow:

and 1.1, flight planning, namely setting parameters such as flight altitude, route, shooting angle, overlapping rate and the like by utilizing professional unmanned aerial vehicle flight planning software according to the scale, terrain and layout of the photovoltaic power station. The method ensures that comprehensive and clear image data are acquired, and avoids shooting dead angles.

And 1.2, image acquisition, namely, operating the unmanned aerial vehicle to fly according to a planned route, and shooting images of the photovoltaic power station from different angles through a carried high-definition camera in the flying process, wherein the equipment such as a photovoltaic module, a bracket, an inverter, a box transformer, a booster station, a weather station and the like and surrounding terrains and buildings are covered.

And 1.3, preprocessing data, wherein the acquired image data may have problems of distortion, noise and the like. The method comprises the steps of performing geometric correction on an image by using an image correction algorithm, removing image deformation caused by a camera lens and a shooting angle, removing noise by using a filtering algorithm, improving image quality, and providing clear and accurate data for subsequent modeling.

2. The detailed processing flow of road network space modeling comprises the following steps:

and 2.1, extracting image features, namely extracting edge contours of the plant area road network and the electrical equipment by using an edge detection algorithm (such as a Canny algorithm), and identifying crossing points of the road network and key feature points of the electrical equipment by using an angular point detection algorithm (such as a Harris angular point detection algorithm).

And 2.2, vectorizing, namely converting the extracted characteristic information into a vector data format. For road network, converting its edge contour into vector line segment, recording the start point, end point coordinates and attribute information of line segment, for electric equipment according to its shape and characteristics, constructing correspondent vector graph, and giving attribute label to describe equipment type and number information.

And 2.3, constructing a topological relation, namely analyzing the connection relation among all line segments in the road network, determining the connectivity and the directivity of the road, and defining the spatial position relation among the electrical equipment, the road network and other equipment to construct a complete topological structure, so that the subsequent spatial analysis and path planning are facilitated.

And 2.4, creating a database, namely creating an electronic map space database by using Geographic Information System (GIS) software according to the vectorization and topological relation construction result. Vector data of the road network and the electrical equipment are stored in a database, classified management is carried out according to different layers, a reasonable index structure is set, and data query and retrieval efficiency is improved.

3. Map modeling detailed processing flow:

and 3.1, model fusion, namely importing a three-dimensional model of power station equipment and a road network space model into three-dimensional modeling software (such as 3ds Max, maya and the like), and performing accurate alignment and fusion according to the actual geographic position and the spatial relationship. The position and the proportion among the models are ensured to be accurate, and a unified three-dimensional scene is formed.

And 3.2, data association, namely establishing association relations with corresponding equipment models or areas for power station construction achievements (such as power station occupied area, installed capacity and the like) and key operation indexes (such as actual power generation capacity, equipment utilization rate and the like) in a three-dimensional digital scene map. Through database connection, the information can be acquired and updated in real time.

And 3.3, visually setting, namely adding an illumination effect to the map, simulating illumination conditions under different time and weather conditions, enhancing the sense of reality of the map, setting materials and textures, giving a realistic appearance to the equipment model and the topography, adding labels and comments, clearly displaying information such as equipment names, numbers, states and the like, and facilitating the user to check and understand.

And 3.4, developing an interactive function by utilizing a programming technology (such as JavaScript combined with a three.js library), realizing zoom, translation and rotation operations of a map by a user, facilitating the user to observe a power station from different angles, and adding a function of checking detailed information by using a click equipment model so as to meet the requirement of the user for deeply knowing the running state of the equipment.

And the anomaly identification unit 4 is used for constructing an anomaly identification model based on deep learning and analyzing potential anomalies of the photovoltaic power station.

In the description of the present invention, the anomaly identification unit 4 includes a model construction training module (not shown), a model early warning analysis module (not shown), a data analysis quantization module (not shown), and a power trend prediction module (not shown).

The model construction training module is used for collecting historical operation data for preprocessing, constructing an anomaly identification model of the long-term and short-term memory network model based on deep learning and carrying out model training.

Specifically, the anomaly identification model structure, construction and training process comprises the following aspects:

1. Model structure

A long-short-term memory network (LSTM) model based on deep learning is employed. LSTM is a special Recurrent Neural Network (RNN) whose core structure includes input gates, forget gates, output gates and memory cells. The input gate determines the input of new information, the forget gate controls the retention or discarding of information in the memory unit, the output gate determines the output value, and the memory unit is used for storing long-term dependent information. The structure can effectively process the long-term dependence problem in the time series data, and is suitable for analyzing the operation data of the photovoltaic equipment changing along with time.

2. Model construction

And 2.1, collecting a large amount of historical operation data of equipment such as an inverter, a photovoltaic module and the like in a photovoltaic power station, wherein the historical operation data comprise physical quantities such as electric quantity (such as input voltage, input current, output active power and the like), temperature, pressure and the like, and equipment operation state identifiers (normal, fault and the like). At the same time, contemporaneous weather data (e.g., light intensity, temperature, wind speed, etc.) is collected, as weather conditions can affect the operation of the device.

And 2.2, preprocessing the data, namely cleaning the collected data to remove abnormal values and missing values. And supplementing the missing values by adopting methods such as mean filling, linear interpolation and the like. And then carrying out normalization processing on the data, mapping all the data to the [0,1] interval, so that different features have the same scale, and improving the training effect of the model.

Dividing the data set, namely dividing the preprocessed data into a training set, a verification set and a test set, wherein the training set, the verification set and the test set are generally divided according to the proportion of 70%, 15% and 15%. The training set is used for model training, the verification set is used for adjusting model super parameters, and the test set is used for evaluating model performance.

3. Model training

And 3.1, selecting a loss function and an optimizer, namely selecting a Mean Square Error (MSE) as the loss function and measuring the difference between the model predicted value and the true value. The optimizer selects an adaptive moment estimation (Adam) optimizer, which can adaptively adjust the learning rate according to the gradient of each parameter, and accelerates the model convergence rate.

And 3.2, training process, namely inputting training set data into the LSTM model according to time step length (for example, one time step every 15 minutes), calculating a predicted value by forward propagation of the model, and then calculating a loss value according to a loss function. And then calculating the gradient through a back propagation algorithm, updating the model parameters, and continuously iterating the training until the loss value on the verification set is no longer reduced or the reduction amplitude is small, so that the model convergence is indicated.

The model early warning analysis module is used for setting input variables and output variables, and predicting the running state of the equipment model at the future moment by using the trained anomaly identification model.

Specifically, the model early warning analysis module is provided with an early warning mechanism, and the principle comprises the following aspects:

1. The input variables comprise real-time operation data (such as input voltage, current, output power and other electric quantities of a plurality of time steps before and at present, and auxiliary variables such as temperature, fan rotating speed and the like) of the inverter, historical fault data (such as fault occurrence time, fault type, equipment operation state before fault and the like), meteorological data (such as illumination intensity, environmental temperature and the like).

2. And the output variable is the probability of failure of the inverter within 2-3 days in the future. The probability of occurrence of faults is larger as the probability approaches to 1, a threshold value (such as 0.5) is set, and early warning is sent out when the prediction probability exceeds the threshold value.

3. And in the analysis process, the model learns the relation and change rule between input variables in the normal operation mode in the training process. In the real-time monitoring stage, the current inverter operation data and the meteorological data are input into a model, and the model predicts the operation state of 2-3 days in the future according to the learned rule. If the prediction result shows that the probability of certain key parameters deviating from the normal range is higher, judging the possible fault type by combining the historical fault data, and sending out early warning when the fault probability exceeds a set threshold value.

And the data analysis and quantization module is used for carrying out differential processing on the time series data, capturing trend characteristics of the operation data by utilizing an autoregressive and moving average model and quantifying the performance degradation degree of the equipment.

In the description of the present invention, the time series data is subjected to differential processing, the trend characteristics of the operation data are captured by utilizing an autoregressive and moving average model, and the performance degradation degree of the quantification equipment comprises:

And S41, converting the acquired real-time operation data of the equipment model into time series data, performing stability test on the time series data, and executing differential processing if the stability threshold is not met.

And S42, setting model parameters of a summation autoregressive moving average model, determining optimal model parameters through a minimum red pool information criterion, fitting time series data by using the optimized summation autoregressive moving average model to obtain a model predicted value, and calculating an error between the model predicted value and a true value.

And S43, analyzing trend items and period items of the time series data based on the fitting result, and quantifying the performance degradation degree of the corresponding electrical equipment of the equipment model according to the change rate of the trend items.

Specifically, the analysis of the trend of the running data of the equipment and the quantification of the performance degradation comprise the following aspects:

1. Analytical algorithm model

An autoregressive integrated moving average model (ARIMA) in time series analysis was used. The ARIMA model smoothes time series data by differential processing and then captures trends and seasonal features in the data using auto-regressive (AR) and Moving Average (MA) models.

2. Quantization process

And (3) carrying out stability test on equipment operation data (such as inverter output power, photovoltaic module current and the like), and carrying out differential processing if the data are unstable.

Parameters (p, d, q) of the ARIMA model (summed autoregressive moving average model), p being the autoregressive order, d being the differential order, q being the moving average order, are determined. The optimal parameters are determined by minimizing the erythro information criterion (AIC) or the Bayesian Information Criterion (BIC).

And (5) fitting the data by using an ARIMA model with determined parameters to obtain a model predicted value. And calculating an error between the predicted value and the true value, and evaluating the fitting effect of the model through error analysis.

And analyzing trend items and period items of the data according to the model fitting result. For example, if the trend term shows a gradual decrease in power and the rate of decrease exceeds a certain threshold, it is determined that degradation of device performance occurs. The degree of performance degradation can be quantified by the slope or percentage of power reduction, e.g., a photovoltaic module has a slope of 0.5%/day power reduction over a month, indicating faster performance degradation.

3. Model formula

In the analysis of the trend of the operational data of the photovoltaic power station equipment and the quantification of performance degradation, a commonly used algorithm model is an autoregressive integral moving average model (ARIMA), and the following calculation formula is as follows:

stationarity test-before the ARIMA model analysis of the equipment operational data, a stationarity test is required on the data, and a common method is a unit root test (such as ADF test). The test regression equation is:

No constant term, no trend term, Δy _t＝ρy_t -1+ Σβ _iΔy_t-i+ε_t (i=1 to p);

a constant term, an no trend term, Δy _t＝α+ρy_t -1+ Σβ _iΔy_t-i+ε_t (i=1 to p);

A constant term, a trend term, Δy _t＝α+β_t+ρy_t -1+ Σβ _iΔyt-i+ε_t (i=1 to p);

Where y _t is time series data, Δy _t＝y_t-y_t -1 represents a first order difference, t is a time trend, and ε _t is a white noise error term. The original assumption for the test is H ₀: ρ=1 (i.e., there is a unit root, the data is not stationary), and if the test statistic is less than the threshold, the original assumption is rejected and the data is considered stationary.

And (3) differential processing: if the data is not stable, differential processing is needed to make the data stable. The d-step difference formula is:

Where Cd++i=d|/(i| (d-i) |) is the number of combinations and d is the differential order. Obtaining a stable sequence after d-order difference

ARIMA (p, d, q) model: after obtaining a stable sequence w _t through difference, an ARIMA (p, d, q) model is constructed, and the expression is as follows:

Wherein Φ (B) =1- Φ1b- Φ2b 2-p is a p-order autoregressive operator, B is a backward-moving operator (By _t＝y_t-1), Φi is an autoregressive coefficient, Θ (B) =1- θ ₁B-θ₂B^2-…-θ_q B q is a q-order moving average operator, θ _i is a moving average coefficient, and ε _t is a white noise sequence.

The autoregressive portion after expansion is w _t＝φ1w_t-1+φ2w_t-2+…+φpw_t-p+ε_t;

the moving average part is epsilon _t＝w_t-θ₁w_t-1-θ₂w_t-2-…-θ_qw_t-q;

model parameter estimation and selection in determining parameters (p, d, q) of the ARIMA model, a common approach is to select the optimal combination of parameters by minimizing the erythro information criterion (AIC) or the Bayesian Information Criterion (BIC).

AIC=-2ln(L)+2k;

BIC=-2ln(L)+kln(n);

Where L is the maximum likelihood estimate of the model, k is the number of parameters to be estimated in the model (k=p+q+1, including constant terms), and n is the number of samples. The (p, d, q) combination that minimizes the AIC or BIC values is selected as the optimal model parameter.

And (3) performance degradation quantification, namely after equipment operation data (such as power data) are fitted and predicted by an ARIMA model, the performance degradation degree can be quantified according to a prediction result. For example, the falling slope of power is calculated to measure performance degradation:

Let the predicted power sequence be The time sequence is t, and the power drop slope S can be calculated by linear regression: The slope S is estimated using a least squares method. If S is negative and its absolute value is greater, this means that the power drops faster and the degradation of the device performance is more severe. The percentage of power reduction may also be calculated to quantify the degree of performance degradation, such as the percentage of power reduction over a period of time t ₁,t₂

Specifically, the photovoltaic module power attenuation trend prediction includes the following aspects:

1. The predicted parameter data comprises historical power data, service life, environment temperature, illumination intensity, accumulated irradiation amount and the like of the photovoltaic module. The historical power data reflects the past performance of the component, the aging degree of the component is reflected by the service life, the power generation efficiency of the component is directly influenced by the ambient temperature and the illumination intensity, and the accumulated irradiation quantity is related to the aging of the component.

2. And predicting the time period, namely predicting the power attenuation trend of 1-5 years in the future according to the data quality and the model performance. Short-term (1-2 years) predictions are mainly used to adjust the power generation plan of a power station and to schedule short-term maintenance in time, and long-term (3-5 years) predictions help the power station plan component replacement cycles and investment budgets. In prediction, the parameter data are input into a trained LSTM model or other regression models (such as Support Vector Regression (SVR)), and the model outputs power predicted values at different time points in the future, so that a power attenuation trend curve is obtained.

And the monitoring alarm unit 5 is used for monitoring the operation parameters of the photovoltaic power station in real time and realizing alarm monitoring.

In the description of the present invention, the monitoring alarm unit 5 includes a status monitoring module (not shown) and an intelligent alarm module (not shown).

The state monitoring module is used for establishing a hierarchical relation model of each equipment model in the photovoltaic power station, performing electric equipment state monitoring and state adjustment based on the hierarchical relation, and performing collaborative optimization with the anomaly identification unit 4 to generate a visual chart of state monitoring data.

In the description of the present invention, the status monitoring module includes a hierarchical relationship sub-module (not shown), a monitoring adjustment sub-module (not shown), a sharing collaboration sub-module (not shown), and a graph analysis sub-module (not shown).

And the hierarchical relation sub-module is used for establishing a hierarchical relation model for sequentially connecting the group string, the combiner box, the inverter, the box transformer and the booster station.

Specifically, it is necessary to preferentially establish a hierarchical relationship model of power generation equipment of a power station, and to clarify the upper and lower relationships of equipment such as a box transformer, an inverter, a string and the like.

The context within a photovoltaic power plant includes the following aspects:

1. the group string and the junction box are connected in parallel, and the group string is the subordinate equipment of the junction box. Each group string gathers the direct current generated by the photovoltaic panel, and transmits the direct current to the junction box, and the junction box gathers and primarily monitors the electric quantity such as current, voltage and the like of the group string.

2. And the junction box is used as a subordinate device of the inverter and transmits the collected direct current to the inverter. The inverter converts direct current into alternating current, and in the process, the inverter monitors the electric parameters input by the bus box to ensure normal operation of the inverter.

3. The inverter and the box transformer are that the alternating current output by the inverter is firstly connected into the box transformer, and the box transformer carries out the rising treatment on the voltage, thereby being convenient for the long-distance transmission of the electric power. The inverter is lower-level equipment of the box transformer, and the box transformer monitors alternating current parameters output by the inverter, so that the stability of power conversion and transmission is ensured.

4. And the electric energy after the box transformer processing is transmitted to the booster station, and the booster station further boosts the voltage to meet the power grid access requirement. The box transformer is the subordinate equipment of the booster station, and the booster station monitors the output electric parameters of the box transformer, so that the safety and stability of the whole power transmission link are ensured.

And the monitoring and adjusting sub-module is used for executing the state monitoring of the electrical equipment based on the upper-level and lower-level relation and adjusting the operation strategy of the upper-level equipment according to the state of the lower-level equipment.

In the description of the present invention, performing electrical device state monitoring based on a superior-subordinate relationship, and adjusting a superior device operation policy according to a subordinate device state includes:

And S51, acquiring real-time operation data of each electrical device obtained through real-time monitoring, and judging that the corresponding electrical device has a fault phenomenon if any one item of data in the real-time operation data does not meet a safety threshold value.

And S52, if the group string current uploaded by the combiner boxes has fault information, calling the combiner boxes to reduce the input power of the corresponding channels, and if the plurality of combiner boxes have fault information, reducing the output power of the inverter and sending alarm information.

And step S53, if the box transformer substation detects that the inverter output has fault information, the tap position of the box transformer substation is adjusted, the transformation ratio is adjusted, and if the fault information is larger than a preset threshold value, the box transformer substation is disconnected from the inverter.

And step S54, if the booster station receives the fault information of the box transformer substation, adjusting the power distribution of other box transformer substation, maintaining the power stability of the photovoltaic power station, and if the fault information of the box transformer substation exceeds the quantity threshold value, reducing the output power of the photovoltaic power station, and cutting off the power supply of a fault area.

Specifically, the detection mode of the fault information in the upper and lower devices includes the following aspects:

1. And detecting the fault of the string, namely installing devices such as a current sensor, a voltage sensor and the like on the string, and collecting electrical parameters such as current, voltage and the like of the string in real time. Comparing the acquired real-time data with a preset normal range, and if the current value exceeds the normal fluctuation range, such as a certain proportion (for example + -20%) lower or higher than the rated current, judging that the current is abnormal, and possibly causing faults. The power output of a string may also be monitored and considered abnormal when the power drops significantly (e.g., 30% below the average power) compared to other strings under the same conditions.

2. And detecting faults of the junction box, namely monitoring parameters such as temperature, communication state and the like of the junction box besides current and voltage of each input group string. By calculating the sum of the currents of each input string and comparing the sum with the output current, if the deviation of the sum and the output current exceeds a certain threshold (such as 5%), an internal line connection problem or a string fault may exist. When the header box temperature is too high (above a set safe temperature, such as 70 ℃), it may indicate that the internal components are radiating abnormally or that an overload condition exists.

3. The inverter fault detection comprises integrating various sensors and monitoring circuits in the inverter, and monitoring the electric quantity of the AC/DC side, such as input voltage, current, output active power, reactive power and the like, and also monitoring the temperature of the inverter, the rotating speed of a fan and the like. And (3) comprehensively analyzing the data, and judging that the inverter possibly has faults when the input voltage exceeds the allowable working range (such as +/-15% of rated voltage) of the inverter or the output power is continuously lower than a certain proportion (such as 70%) of the rated power under the normal illumination condition, and combining the conditions of abnormal temperature rise and the like.

4. And (3) detecting the fault of the box transformer, namely installing a current transformer and a voltage transformer on the high-low voltage side of the box transformer to monitor current and voltage, monitoring the internal gas state through a gas relay, and monitoring the oil temperature through a thermometer. When the three-phase current imbalance exceeds a specified value (e.g., 15%), or the oil temperature exceeds a permissible value (e.g., 85 ℃), the gas relay is operated, which means that the tank transformer may be faulty.

5. And the booster station utilizes a relay protection device to monitor the changes of the parameters in real time by monitoring the electrical parameters of various electrical equipment, such as bus voltage, line current, power factor and the like. When the voltage deviation exceeds a specified range (such as +/-10% rated voltage) or abnormal conditions such as short circuit current occur, the relay protection device rapidly acts, and the fault of the booster station is judged.

The manner in which the superior device adjusts the operation policy according to the state of the inferior device includes the following aspects:

1. When the inverter receives abnormal information of a certain group of series currents uploaded by the bus box, the inverter firstly tries to reduce the input power of a corresponding channel of the bus box, so that the inverter is prevented from being excessively impacted by abnormal currents. If a plurality of junction boxes are abnormal, the inverter appropriately reduces the overall output power to protect its own components. Meanwhile, the inverter can send alarm information to the monitoring system to prompt operation and maintenance personnel to timely process the fault of the junction box.

2. The box transformer is based on an inverter state adjustment strategy in that if the box transformer detects an abnormal inverter output, such as excessive power fluctuation or unbalanced three-phase current, the box transformer adjusts its tap position and changes the transformation ratio to stabilize the output voltage. If the inverter has serious faults, the box transformer can cut off the connection with the inverter, prevent the faults from expanding to the whole power grid and upload fault information to the booster station.

3. And the booster station adjusts according to the box-type state adjustment strategy, namely, after the booster station receives fault information of the box-type transformer substation, the booster station adjusts according to the severity of the fault and the running condition of the power grid. If the individual box transformer is slightly faulty, the booster station can adjust the power distribution of other box transformers to maintain the power output of the whole power station stable. If the fault is serious, the booster station can be coordinated with the power grid dispatching center, so that the overall output power of the power station is reduced, even the power supply of a fault area is cut off, and the safety of the power grid is ensured.

Further, the detection device and the communication device facing the upper and lower devices include:

1. The detection equipment comprises various sensors, such as a current sensor (for measuring the current), a voltage sensor (for measuring the voltage), a temperature sensor (for monitoring the temperature of the equipment), a gas relay (for detecting the gas state in the box transformer) and the like, and the sensors are responsible for collecting various operation parameters of the equipment. And the relay protection device is used for monitoring abnormal changes of the electrical parameters, and rapidly acts when the parameters exceed a set threshold value to protect the safety of equipment.

2. The communication equipment mainly comprises optical fibers, wireless communication modules (such as a 4G/5G module, a LoRa module and the like), an industrial Ethernet switch and the like. The optical fiber has the characteristics of high transmission speed and high anti-interference capability, and is used for long-distance and high-speed data transmission and connection with important equipment and monitoring centers in a power station. The wireless communication module is suitable for some areas with difficult wiring, and wireless data transmission between equipment and a monitoring system is realized. The industrial Ethernet switch is used for constructing a local area network in the station to realize data exchange and communication between the devices.

And the sharing coordination sub-module is used for synchronizing real-time operation data and monitoring data of the electrical equipment of each level to the abnormality recognition unit 4, acquiring a trend prediction result output by the abnormality recognition model, and executing response control between the upper and lower devices according to faults or abnormalities of the electrical equipment.

Specifically, the association of the shared collaboration sub-module with the abnormality recognition model within the abnormality recognition unit 4 includes the following aspects:

1. And the data sharing is that a large amount of operation data acquired by the detection equipment is not only the basis of traditional fault detection, but also the input data source of the abnormality identification model. The anomaly identification model utilizes the rich data to mine potential fault modes and rules through algorithms such as deep learning and the like, and provides more accurate judgment for fault detection.

2. The traditional threshold-based fault detection mode may have limitations on the judgment of some complex faults or early faults. The anomaly identification model can discover some hidden fault signs through comprehensive analysis of the data. The two are combined, so that the accuracy and timeliness of fault diagnosis can be improved. When detecting string faults, the anomaly identification model can analyze the change trend of string current and voltage along with time and correlate with other environmental factors, predict possible faults in advance and complement with a detection mode based on a threshold value.

3. And the abnormal recognition model can provide more scientific basis for the adjustment of the operation strategy of the superior equipment according to the historical operation data and the fault data of the equipment. When the inverter adjusts the operation strategy according to the state of the combiner box, the abnormal recognition model can analyze the optimal adjustment scheme under different fault conditions, so that the inverter is helped to more reasonably adjust the power output, and the influence on the overall operation of the power station is reduced.

4. And controlling from top to bottom, namely, the booster station sends a power adjustment instruction to the box transformer according to the power grid demand. The box transformer adjusts output according to the instruction, and monitors the working state of the lower inverter to ensure that the output meets the requirement. The inverter adjusts the running parameters of the inverter according to the box transformer instruction, and correspondingly controls the lower-stage junction box and the group string, so that the stability and the reliability of power output are ensured.

Specifically, the chart analysis sub-module draws a visual chart according to the electrical equipment operation data, including the following aspects:

1. And collecting and arranging data, namely continuously collecting power generation data of each device of the photovoltaic power station, including power, electric quantity and the like, and environment data, such as illumination intensity, temperature and the like. And (3) sorting the collected data, removing abnormal values and error data, and ensuring the accuracy and the integrity of the data.

2. And selecting an analysis algorithm, namely selecting a proper data analysis algorithm according to the characteristics of the data and the analysis purpose. The conventional method comprises a moving average method for calculating the average value of data in a certain time period, smoothing the data fluctuation and showing the approximate trend of the power generation trend, and a time sequence analysis method for predicting future power generation data by considering the time sequence and trend characteristics of the data, so as to provide decision basis for power station operation.

3. Chart drawing the analyzed data is plotted using a specialized data visualization tool, such as Excel, python's Matplotlib library, etc. And selecting a proper chart type, for example, the line graph can clearly show the change trend of the generated power along with the time, and the histogram can visually compare the generated power in different time periods. Elements such as titles, coordinate axis labels, legends and the like are added to the chart, so that the chart is easy to understand and read.

In the description of the present invention, the intelligent warning module includes a warning analysis sub-module (not shown), a threshold optimization sub-module (not shown), a warning support sub-module (not shown), and an analysis assistance sub-module (not shown).

And the early warning analysis sub-module is used for collecting monitoring data of each electrical device in real time, judging and analyzing whether each electrical device has a fault through threshold judgment, and triggering early warning and reminding after the fault is found.

Specifically, the real-time monitoring, early warning reminding and equipment maintenance aiming at different electrical equipment and corresponding equipment types thereof comprise the following aspects:

1. And (5) alarming over temperature of the inverter:

And judging and analyzing, namely installing a temperature sensor on the inverter, and monitoring the temperature of key components (such as a power module) in the inverter in real time. By setting a temperature threshold value for normal operation of the inverter, which is assumed to be 80 ℃, when the real-time temperature acquired by the sensor exceeds the threshold value, the system judges an over-temperature condition and triggers an alarm. This is based on long-term operation experience of the inverter and on thresholds determined by technical parameters provided by the equipment manufacturer, and excessive temperatures can accelerate electronics aging, reduce inverter efficiency, and even cause failure.

The early warning function is to inform operation and maintenance personnel in advance that the temperature of the inverter is abnormal, so that the damage of the inverter caused by the continuous temperature rise is avoided, the photovoltaic power generation efficiency is influenced, and the power failure loss is reduced.

2. And (3) alarming unbalanced three-phase current of the box transformer:

And judging and analyzing, namely installing a current sensor on a three-phase line of the box transformer substation, and collecting three-phase current values in real time. Calculating the unbalance degree of the three-phase current, and giving an alarm when the unbalance degree exceeds a set threshold (such as 15%). The unbalance is calculated based on the difference between the average value and the maximum value of the three-phase current, and this threshold is determined based on the design criteria and actual operation experience of the tank transformer. The unbalance of three-phase current can increase transformer loss, affect the service life of the transformer, and can cause abnormal operation of equipment such as a motor.

The early warning function is that operation and maintenance personnel can timely find potential problems in the operation of the box transformer substation, measures are taken to adjust load distribution, the box transformer substation is prevented from being damaged due to unbalanced three-phase current, and the stability of power transmission is ensured.

3. Photovoltaic string power anomaly reduction warning:

And judging and analyzing, namely comparing the output power of each photovoltaic group string with the historical contemporaneous power data and the power of the adjacent group string by monitoring. If the power of a certain group of strings is lower than a set power threshold (such as lower than 80% of the historical average power) under the conditions that the illumination condition is normal and the inverter works normally, the system triggers an alarm. The judging mode comprehensively considers the environmental factors and the self performance of the equipment, and can accurately identify the abnormal situation of the group string power. Possible causes include photovoltaic panel shielding, assembly damage, poor line contact, etc.

The early warning function is to help operation and maintenance personnel to quickly locate the string with abnormal power, to timely troubleshoot the fault, to recover the power generation efficiency of the photovoltaic power station and to reduce the power generation loss.

4. An inverter over-temperature warning remedial measure:

And (3) checking the heat radiation system, namely after the operation and maintenance personnel receive the alarm, firstly checking whether the heat radiation fan of the inverter normally operates and whether the ventilation opening is blocked. If the fan is out of order, the fan is replaced in time, and if the vent is blocked, sundries in the vent are cleaned, so that good heat dissipation conditions are ensured.

And the load is reduced, namely, the load of the inverter is properly reduced and the working power is reduced when the condition allows, so that the heating value is reduced. The access of the group string can be temporarily reduced by adjusting the power output strategy of the power station, and the normal operation is gradually restored after the temperature is restored to be normal.

And monitoring the temperature change, namely continuously monitoring the temperature change of the inverter after the measures are taken, and ensuring that the temperature gradually decreases and returns to be within a normal range. If the temperature is not controlled effectively, the professional technician is contacted for further maintenance.

5. And (3) a box transformer three-phase current unbalance warning and remedying measure:

and (3) adjusting load distribution, namely, reallocating the three-phase load according to the load condition of box transformer connection by operation and maintenance personnel, so as to balance the three-phase load as much as possible. For example, a partial single phase load is transferred from a phase with a larger current to a phase with a smaller current to reduce the imbalance of the three phase currents.

Checking line connection, namely checking the connection part of the box change-in line and checking whether loosening, oxidization and the like exist. If the connection is found to be poor, fastening and processing are timely carried out, so that good line contact is ensured, contact resistance is reduced, and unbalance of current caused by line problems is avoided.

And (3) monitoring the current change, namely closely monitoring the change condition of three-phase current of the box transformer after load adjustment and line inspection are completed, and ensuring that the unbalance degree is recovered to be in a normal range. The box transformer substation is periodically inspected, similar problems are found and processed in time, and stable operation of the box transformer substation in a variable length period is ensured.

6. Photovoltaic string power anomaly reduction warning remedial action:

checking the shielding condition, namely rapidly going to the position of the alarm group string and checking whether shielding objects such as leaves, dust, bird droppings and the like exist on the surface of the photovoltaic panel. If the shielding exists, timely cleaning the shielding object, and ensuring that the photovoltaic panel can fully receive illumination.

And detecting the components and the circuits, namely detecting the photovoltaic components by using professional detection equipment and judging whether the components are damaged. At the same time, it is checked whether the connection of the string lines is firm, whether an open or short condition exists. If the damaged component is found, the damaged component is replaced in time, and if the circuit has a problem, the circuit is repaired or replaced.

Recording and analyzing data, namely recording the occurrence time, the occurrence reason and the processing process of the fault after the fault is processed. And analyzing the related data to summarize the occurrence rule of faults so as to take preventive measures later and improve the overall operation reliability of the photovoltaic power station.

And the threshold optimization sub-module is used for analyzing the change rules of the operation of the electrical equipment under different working conditions and environments through the anomaly identification model and dynamically adjusting and judging the threshold.

Specifically, parameters obtained by intelligent alarm detection are closely related to the abnormal recognition model and are not independently operated. The two supplement each other, and the stable operation of the photovoltaic power station is ensured together. The intelligent alarm provides basic data for the abnormal recognition model, and the abnormal recognition model improves the accuracy and the foresight of the intelligent alarm. The threshold optimization sub-module establishes data communication with the anomaly identification model, and the implementation of threshold optimization and adjustment comprises the following aspects:

1. The data provides that parameters of intelligent alarm detection, such as inverter temperature, box transformer three-phase current, photovoltaic string power and the like, are important data sources of an abnormality identification model. These real-time monitoring data provide a large number of samples for model training, enabling it to learn the characteristics of the device in normal and abnormal conditions. Through inverter temperature data accumulated for a long time, an abnormality recognition model can construct a more accurate temperature change trend model to recognize potential over-temperature risks, and the potential over-temperature risks are not only dependent on a preset threshold value.

2. And (3) optimizing threshold judgment, namely optimizing the threshold of the intelligent alarm by the abnormal recognition model according to historical data and a complex algorithm. The thresholds of the traditional intelligent alarms are mostly set based on experience and equipment standards, and the flexibility is lacking. The abnormal recognition model can analyze the change rule of the equipment parameters under different working conditions and environments, and dynamically adjust the threshold value. Under different seasons and illumination intensity, the judging threshold value of the abnormal power of the photovoltaic string can be correspondingly adjusted, the alarm accuracy is improved, and false alarm and missing report are reduced.

And the early warning support sub-module is used for acquiring a trend prediction result of the anomaly identification model and fault monitoring information of the electrical equipment, and triggering early warning reminding and maintenance response in advance.

Specifically, the anomaly identification model can discover potential faults in advance by analyzing the association and trend among multiple parameters, and provides more prospective information for intelligent warning. By combining parameters such as the temperature, the current, the power and the like of the inverter, the model can predict the impending over-temperature fault, so that the intelligent alarm gives out early warning in advance, more abundant processing time is given to operation and maintenance personnel, and compared with the method that the alarm is triggered by simply depending on the real-time parameter exceeding a threshold value, the damage of equipment and the reduction of the power generation efficiency can be prevented better.

Specifically, after the intelligent alarm is triggered, the abnormal recognition model can assist in analyzing the fault cause. For the unbalanced warning of the three-phase current of the box transformer, the model can synthesize other related parameters and historical data, judge whether the fault is caused by a load problem, a line fault or a defect of equipment, provide a direction for operation and maintenance personnel to rapidly check the fault, and improve the fault processing efficiency.

In summary, by means of the technical scheme, the invention aims to solve the problems of incomplete information and unsophisticated monitoring of the traditional photovoltaic power station monitoring system, and builds a comprehensive power station information display platform through three-dimensional digital twin modeling and real-time data hooking, so that the dynamic monitoring of power station equipment can be realized, the difficulties are effectively solved, the efficiency and the accuracy of monitoring the photovoltaic power station are improved, the defects of the traditional photovoltaic power station monitoring system in information acquisition and data analysis are overcome, and solid guarantee is provided for safe and efficient operation of the power station. The invention realizes the omnibearing digital monitoring of the photovoltaic power station, acquires the attribute of the multidimensional electrical equipment through the model construction unit, constructs an equipment model, then connects the model with real-time operation data in a hanging way, efficiently manages the data by using a distributed storage technology, ensures the timeliness and the accuracy of information transmission, realizes the dynamic and visual display of the states of all the equipment by the three-dimensional overview unit, ensures that operation staff can quickly know the overall operation condition of the power station, captures potential faults in advance by the abnormal recognition unit based on deep learning, and can send early warning at the first time by matching with the real-time warning monitoring unit, thereby reducing the risk of accidents. The invention not only greatly improves the intellectualization and safety of the operation of the power station, but also provides scientific basis for the operation and maintenance decision of the power station, thereby being beneficial to reducing the maintenance cost and prolonging the service life of equipment.

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.

Claims

1. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station, characterized by comprising:

A model building unit, used to obtain multi-dimensional electrical equipment attribute parameters and build an equipment model;

A mounting storage unit is used to mount the device model to the real-time operation data and perform distributed storage on the real-time operation data;

The three-dimensional overview unit is used to construct a three-dimensional model of the photovoltaic power station and dynamically display the real-time operating status of each equipment model of the photovoltaic power station;

Anomaly recognition unit, which is used to build an anomaly recognition model based on deep learning and analyze potential anomalies in photovoltaic power plants;

The monitoring and alarm unit is used to monitor the operating parameters of the photovoltaic power station in real time and realize alarm monitoring.

2. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 1, characterized in that the mounting storage unit comprises:

Dynamic hookup module, used to create data interfaces, connect data interactions between device models and real-time operation data, and build a hierarchical communication architecture to monitor data hookup and updates in real time;

The storage management module is used to adopt distributed storage technology to disperse and store the real-time operation data, meteorological data and model data of the photovoltaic power station to multiple nodes, and introduce redundant backup mechanism and erasure code technology to maintain the long-term availability and integrity of the data.

3. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 1, characterized in that the three-dimensional overview unit comprises:

The 3D modeling module is used to combine oblique photography with vector modeling to build a 3D model of a photovoltaic power station, supporting real-time data-driven dynamic display of the model;

The road network space module is used to digitally reconstruct the factory road network and electrical equipment based on drone aerial image data to form an electronic map space database;

The map overview module is used to build a three-dimensional digital scene map to display the real-time operating status of various types of equipment models in the photovoltaic power station.

4. According to claim 1, a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station is characterized in that the abnormality identification unit comprises:

Model building and training module, which is used to collect historical operation data for preprocessing, build an abnormality recognition model based on the long short-term memory network model of deep learning, and perform model training;

The model early warning analysis module is used to set input variables and output variables, and use the trained abnormality recognition model to predict the operating status of the equipment model in the future;

Data analysis and quantification module, used to perform differential processing on time series data, using autoregression and sliding average models to capture the trend characteristics of operating data and quantify the degree of equipment performance degradation;

The power trend prediction module is used to obtain the historical photovoltaic data of the photovoltaic modules, use the anomaly recognition model to predict the power attenuation trend of the photovoltaic modules in the future time period, draw the power attenuation trend curve, and respond to equipment maintenance measures according to the trend prediction results.

5. According to claim 4, a three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station is characterized in that the differential processing of the time series data, the use of autoregressive and sliding average models to capture the trend characteristics of the operating data, and the quantification of the degree of equipment performance degradation include:

The real-time operation data of the collected equipment model is converted into time series data, and the time series data is tested for stationarity. If the stationarity threshold is not met, differential processing is performed;

The model parameters of the summed autoregressive moving average model are set, the optimal model parameters are determined by minimizing the Akaike information criterion, the optimized summed autoregressive moving average model is used to fit the time series data, the model prediction value is obtained, and the error between the model prediction value and the true value is calculated;

Based on the fitting results, the trend term and period term of the time series data are analyzed, and the performance degradation degree of the electrical equipment corresponding to the equipment model is quantified according to the change rate of the trend term.

6. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 1, characterized in that the monitoring alarm unit comprises:

The status monitoring module is used to establish a hierarchical relationship model of each device model in the photovoltaic power station, perform electrical equipment status monitoring and status adjustment based on the hierarchical relationship, and coordinate optimization with the abnormality identification unit to generate a visual chart of the status monitoring data;

The intelligent alarm module is used to use sensors to collect monitoring data of various electrical equipment in real time, determine whether the equipment model triggers an alarm response, and analyze the cause of the fault.

7. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 6, characterized in that the status monitoring module comprises:

The hierarchical relationship submodule is used to establish a hierarchical relationship model in which strings, combiner boxes, inverters, box transformers and booster stations are connected in sequence;

The monitoring and adjustment submodule is used to perform electrical equipment status monitoring based on the upper and lower level relationships, and adjust the upper level equipment operation strategy according to the lower level equipment status;

The shared collaboration submodule is used to synchronize the real-time operation data and monitoring data of electrical equipment at each level to the abnormality identification unit, obtain the trend prediction results output by the abnormality identification model, and execute response control between upper and lower level equipment according to the faults or abnormalities existing in the electrical equipment;

The chart analysis submodule is used to draw visual charts based on the operating data of electrical equipment.

8. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 7, characterized in that the electrical equipment status monitoring based on the superior-subordinate relationship and adjusting the superior equipment operation strategy according to the subordinate equipment status include:

Acquire the real-time operation data of each electrical device obtained through real-time monitoring. If any item of the real-time operation data does not meet the safety threshold, it is determined that the corresponding electrical device has a fault phenomenon;

If the string current uploaded by the combiner box has fault information, the combiner box is called to reduce the input power of the corresponding channel. If multiple combiner boxes have fault information, the inverter output power is reduced and an alarm message is sent;

If the box-type transformer detects fault information on the inverter output, the box-type transformer tap position is adjusted to adjust the transformation ratio. If the fault information is greater than a preset threshold, the box-type transformer is disconnected from the inverter.

If the booster station receives fault information from a box transformer, it will adjust the power distribution of other box transformers to maintain the power stability of the photovoltaic power station. If the box transformer fault information exceeds the quantity threshold, the output power of the photovoltaic power station will be reduced and the power supply to the fault area will be cut off.

9. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 6, characterized in that the intelligent alarm module comprises:

The early warning analysis submodule is used to collect monitoring data of various electrical equipment in real time, judge and analyze whether each electrical equipment has a fault through threshold judgment, and trigger an early warning reminder after a fault is found;

The threshold optimization submodule is used to analyze the changing rules of electrical equipment operation under different working conditions and environments through the abnormal recognition model, and dynamically adjust the judgment threshold;

The early warning support submodule is used to obtain the trend prediction results of the abnormal recognition model and the fault monitoring information of the electrical equipment, and trigger early warning reminders and maintenance responses;

The analysis assistance submodule is used to match the fault cause and output the fault maintenance direction.

10. A three-dimensional digital intelligent monitoring system suitable for a photovoltaic power station according to claim 1 or 2, characterized in that the equipment model includes: a power station model, a booster station model, a box transformer model, an inverter model, a junction box string model, a light resource model and other equipment models.