Distributed maintenance method and system for photovoltaic power station

Abstract

The invention discloses a distributed maintenance method and system for a photovoltaic power station, and belongs to the technical field of intelligent operation and maintenance of the photovoltaic power station, and the method comprises the steps: carrying out the key node recognition of distributed design information; control module deployment is carried out on N photovoltaic power station key nodes in sequence, and each node control module in N networking node control module sets comprises a data sensing unit, a data processing unit and a data transmission unit. N key node working data streams are obtained, and distributed fault analysis is carried out on the N key node working data streams; integrating the N key node fault analysis parameter sets to a control cloud center through a data transmission unit for maintenance analysis; and carrying out distributed maintenance and feedback adjustment optimization on the target photovoltaic power station. The technical problems of large maintenance blind area, fault processing delay, high operation and maintenance cost and high fault risk caused by lack of real-time monitoring and distributed fault diagnosis means for the running state of the photovoltaic power station in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation and maintenance of photovoltaic power plants, and in particular to a distributed maintenance method and system for photovoltaic power plants. Background Art

[0002] As global climate change and environmental problems become increasingly severe, promoting energy transformation and developing renewable energy has become a global consensus. Photovoltaic power generation, as a clean and efficient way to utilize renewable energy, has developed rapidly around the world. The installed capacity of photovoltaic power stations continues to rise, the scale is getting larger and the distribution is becoming more extensive.

[0003] At present, the operation and maintenance technologies of photovoltaic power stations mainly include manual inspection, regular maintenance, centralized monitoring and other methods. Manual inspection is inefficient and difficult to cover large-scale, distributed power stations; regular maintenance has problems such as long maintenance cycles and difficulty in timely detection of faults; although the centralized monitoring system can monitor the overall operation status of the power station, it is difficult to perform fine management and fault diagnosis for each component or key node. In recent years, with the development of technologies such as the Internet of Things, big data, and artificial intelligence, some new operation and maintenance technologies have begun to be applied to photovoltaic power stations, but these technologies still have some limitations. Therefore, there is an urgent need to develop a more intelligent and efficient photovoltaic power station operation and maintenance method to improve the operation efficiency of the power station and reduce the operation and maintenance costs. Summary of the invention

[0004] The present application provides a distributed maintenance method and system for photovoltaic power stations, aiming to solve the technical problems in the prior art of lacking real-time monitoring of the operating status of photovoltaic power stations and distributed fault diagnosis means, which leads to large maintenance blind spots, delayed fault handling, high operation and maintenance costs and high fault risks.

[0005] In view of the above problems, the present application provides a distributed maintenance method and system for a photovoltaic power station.

[0006] In a first aspect disclosed in the present application, a distributed maintenance method for a photovoltaic power station is provided, the method comprising acquiring distributed design information of a target photovoltaic power station, identifying key nodes of the distributed design information, and obtaining N key nodes of the photovoltaic power station; sequentially deploying control modules on the N key nodes of the photovoltaic power station to obtain N networking node control modules, wherein each node control module in the N networking node control modules includes a data sensing unit, a data processing unit, and a data transmission unit; acquiring N key node working data streams through the data sensing unit, performing distributed fault analysis on the N key node working data streams based on the data processing unit, and obtaining N key node fault analysis parameters; integrating the N key node fault analysis parameter sets into a control cloud center through the data transmission unit for maintenance analysis, and obtaining N key node fault maintenance plans; and performing distributed maintenance and feedback adjustment optimization on the target photovoltaic power station based on the N key node fault maintenance plans.

[0007] Another aspect disclosed in the present application provides a distributed maintenance system for a photovoltaic power station, the system comprising a key node identification module, used to obtain distributed design information of a target photovoltaic power station, perform key node identification on the distributed design information, and obtain N key nodes of the photovoltaic power station; a control module deployment module, used to sequentially deploy control modules on the N key nodes of the photovoltaic power station, and obtain N networking node control modules, wherein each node control module in the N networking node control modules includes a data sensing unit, a data processing unit, and a data transmission unit; a distributed fault analysis module, used to collect and obtain N key node working data streams through the data sensing unit, perform distributed fault analysis on the N key node working data streams based on the data processing unit, and obtain N key node fault analysis parameters; a maintenance analysis module, used to integrate the N key node fault analysis parameter sets into a control cloud center through the data transmission unit for maintenance analysis, and obtain N key node fault maintenance plans; an adjustment and optimization module, used to perform distributed maintenance and feedback adjustment and optimization on the target photovoltaic power station based on the N key node fault maintenance plans.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By adopting a technical solution of obtaining distributed design information of photovoltaic power stations to identify key nodes, deploying control modules including data sensing units, data processing units and data transmission units at key nodes, collecting working data streams for distributed fault analysis, integrating fault analysis parameters to the cloud center to generate maintenance plans and implement feedback optimization, the technical problems of lack of real-time monitoring and distributed fault diagnosis means for the operating status of photovoltaic power stations in the existing technology, which leads to large maintenance blind spots, delayed fault handling, high operation and maintenance costs and failure risks, are solved. The technical effects of accurately identifying photovoltaic power station faults, quickly formulating maintenance plans, significantly improving operation and maintenance efficiency and reducing downtime are achieved.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of a distributed maintenance method for a photovoltaic power station is provided for an embodiment of the present application;

[0012] Figure 2 A structural schematic diagram of a distributed maintenance system for a photovoltaic power station is provided for an embodiment of the present application.

[0013] Explanation of the reference numerals: key node identification module 11 , control module deployment module 12 , distributed fault analysis module 13 , maintenance analysis module 14 , adjustment and optimization module 15 . DETAILED DESCRIPTION

[0014] The overall idea of ​​the technical solution provided by this application is as follows:

[0015] The embodiment of the present application provides a distributed maintenance method and system for photovoltaic power stations. By obtaining the distributed design information of the photovoltaic power station to identify key nodes, and deploying control modules with integrated data perception, processing and transmission functions at the key nodes, real-time collection of node working data and distributed fault analysis are realized. Combined with the centralized computing power of the cloud center, integrated fault analysis parameters are used to generate maintenance plans and implement feedback optimization, thereby improving the operation monitoring efficiency, fault diagnosis accuracy and maintenance response speed of the photovoltaic power station and reducing operation and maintenance costs.

[0016] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0017] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a distributed maintenance method for a photovoltaic power station, the method comprising:

[0018] Step S100: obtaining distribution design information of a target photovoltaic power station, identifying key nodes of the distribution design information, and obtaining N key nodes of the photovoltaic power station.

[0019] Specifically, the distribution design information of the target PV power station refers to the overall layout information of the target PV power station, including the spatial distribution and connection methods of PV modules, inverters, power distribution equipment, transmission lines, etc. Key node identification refers to selecting nodes that have a significant impact on operating efficiency, failure risk or maintenance requirements from all nodes of the PV power station. The N key nodes of the PV power station represent the N nodes with higher priority that require special attention and maintenance after identification. The specific value of N is determined by the scale and complexity of the PV power station and the results of the algorithm screening, where N is a positive integer greater than or equal to 1.

[0020] First, layout data is extracted from the distribution design information of the target PV power station to generate a basic data set, which includes the geographical location, equipment type and connection relationship of the nodes. Then, these layout data are constructed into a three-dimensional model of the PV power station. Then, based on the three-dimensional model, the distribution design information of the power station is gridded. Then, the maintenance importance of each grid node is evaluated by the criticality analysis method. Finally, the nodes are screened based on the criticality coefficient. The screening threshold is set, all grid nodes are sorted and N key nodes of the PV power station are screened.

[0021] Through this step, the key nodes of the PV power station can be efficiently identified, the high cost and low efficiency of comprehensive maintenance can be avoided, and the ability of fault diagnosis and optimized resource allocation can be enhanced.

[0022] Step S200: Control modules are deployed sequentially on the N key nodes of the photovoltaic power station to obtain N networking node control modules, wherein each node control module in the N networking node control modules includes a data sensing unit, a data processing unit and a data transmission unit.

[0023] Specifically, the control module refers to an intelligent hardware device installed on a key node, which is used for real-time monitoring, data collection, fault analysis, remote control and other functions. It usually integrates data perception, processing, transmission and other functions, and can be linked with other nodes or systems. The networking node control module refers to the role of the control module in the entire photovoltaic power station network, through which the key nodes of the photovoltaic power station are controlled, monitored and data transmitted. The networking node is a summary of the various control modules in the network. The data perception unit is responsible for collecting environmental data and equipment status data, usually composed of sensors or measuring instruments. It can sense key parameters such as current, voltage, temperature, and radiation intensity in real time. The data processing unit processes and analyzes the data collected by the data perception unit, including data cleaning, fault detection, parameter calculation and prediction. It is the "brain" in the control module, which can judge the operation status of the power station according to the algorithm and make decisions. The data transmission unit is responsible for transmitting the perceived real-time data and processing results to the central control system or other nodes through the network. It usually includes wireless communication modules (such as Wi-Fi, ZigBee, 4G / 5G) or wired communication modules (such as Ethernet).

[0024] First, according to the N key nodes of the photovoltaic power station, the control modules are deployed on these nodes in turn according to the preset plan. The deployed control modules can be responsible for monitoring the operating status of the inverter, including parameters such as voltage, current, and temperature. Among them, the data perception unit of each control module will be equipped with different types of sensors, and the sensors will be selected according to the actual needs of the node. These data perception units will continuously monitor the environment and equipment status and generate real-time data streams. The collected raw data will be transmitted to the data processing unit, which will perform real-time data analysis according to the preset algorithm. The processed data will be transmitted to the central control system or other modules that need to share data through the data transmission unit.

[0025] By deploying control modules at key nodes, distributed monitoring and intelligent management of photovoltaic power stations are realized. The operating status of each key node can be monitored in real time, faults can be quickly discovered, and remote control and maintenance can be performed, which improves the operating efficiency and fault response speed of the power station.

[0026] Step S300: Acquire N key node working data streams through the data perception unit, perform distributed fault analysis on the N key node working data streams based on the data processing unit, and obtain N key node fault analysis parameters.

[0027] Specifically, the working data stream refers to the data stream collection collected from each key node in real time, which usually contains a large amount of information related to the operating status of the photovoltaic power station. Distributed fault analysis refers to the use of multiple control modules and nodes to detect and analyze the operating conditions of each key node using the collected data stream. Fault analysis parameters are the results obtained after analysis, usually including the fault type of each key node, the severity of the fault, the time of fault occurrence, possible causes and other information.

[0028] The key nodes of each photovoltaic power station collect field working data in real time through the data sensing unit on its control module. These data are packaged and transmitted to the data processing unit to form a data stream. The data processing unit receives working data streams from multiple nodes and analyzes these data. Through preset algorithms or models (such as machine learning, timing analysis, anomaly detection, etc.), the data processing unit will determine the working status of each node and identify whether there is a fault. Suppose the data processing unit finds that the output power of a certain inverter has dropped significantly and the voltage fluctuation exceeds the preset threshold by comparing historical data and real-time data. Through the fault analysis model, it will be determined that the inverter has an electrical fault, such as a short circuit or an internal connection fault. After completing the fault analysis, the data processing unit will generate fault analysis parameters. These parameters include but are not limited to the type of fault, the time of occurrence, the severity of the fault, possible causes, etc.

[0029] Through this distributed fault analysis method based on data sensing unit working data flow collection and data processing unit, real-time monitoring and intelligent fault diagnosis of key nodes of photovoltaic power stations can be achieved. Due to the use of distributed analysis, data between each node is processed independently, the analysis efficiency is high and the accuracy is high, which improves the timeliness and accuracy of fault detection.

[0030] Step S400: Integrate the N key node fault analysis parameter sets into the control cloud center through the data transmission unit for maintenance analysis to obtain N key node fault maintenance solutions.

[0031] Specifically, the control cloud center is a centralized data processing and analysis platform that receives data from various photovoltaic power station nodes and conducts complex analysis, prediction and maintenance decisions. The cloud center usually relies on big data technology, artificial intelligence algorithms and efficient data storage capabilities.

[0032] First, the data transmission unit is responsible for integrating the N key node fault analysis parameter sets previously obtained by the data processing unit. The key nodes of each photovoltaic power station package the analysis results and fault diagnosis parameters into standardized data packets through the control module, and transmit the data to the control cloud center using wireless communication or wired network. In the cloud center, the received fault analysis parameters will be further processed. The cloud platform conducts a comprehensive analysis based on these data and generates corresponding fault maintenance plans. Specifically, in the cloud center, a comprehensive analysis is conducted in combination with historical data, operation models and expert rule bases. The cloud platform uses machine learning algorithms to identify fault modes and formulates maintenance plans based on optimization algorithms. The plan content includes fault type, priority, resource allocation suggestions, etc. Through parallel computing on big data platforms (such as Hadoop and Spark), maintenance strategies can be quickly generated and pushed to each key node for execution, ensuring efficient maintenance and fault handling of photovoltaic power stations.

[0033] This method can effectively integrate fault analysis data from various key nodes, and perform intelligent analysis and maintenance plan generation through the cloud center, thereby achieving early warning and timely repair of faults, and improving the operating efficiency and reliability of photovoltaic power stations.

[0034] Step S500: performing distributed maintenance and feedback adjustment optimization on the target photovoltaic power station based on the N key node failure maintenance solutions.

[0035] Specifically, distributed maintenance means that the control module of each node performs corresponding maintenance tasks according to the N key node fault maintenance plans generated by the control cloud center.

[0036] The N key node fault maintenance plans generated by the cloud center will be sent to the control modules of each key node through the data transmission unit. After receiving the plan, the control module of each node starts the local execution unit, such as adjusting the inverter parameters, triggering an alarm signal, starting the backup equipment, or shutting down the faulty PV modules. At each key node, the control module independently completes the relevant maintenance tasks according to the maintenance plan. The control module of each node completes the task through local sensors and actuators. For example, the angle adjustment of the PV module can be achieved through the motor control module, and the inverter fault switching can be completed through the circuit breaker control module.

[0037] After performing maintenance tasks, each control module collects the operation feedback data of the node in real time, such as whether the fault has been repaired, whether the voltage and current have returned to normal, etc. These data are transmitted back to the control cloud center through the data transmission unit, and the cloud center analyzes the maintenance effect based on the feedback data. If the feedback data shows that the maintenance effect of some nodes is not good, the cloud center will re-optimize the maintenance plan and send it again.

[0038] Through distributed maintenance and feedback adjustment optimization, photovoltaic power stations achieve accurate and efficient maintenance and operation optimization. Distributed maintenance reduces the delay of centralized operations, increases the speed of node fault repair, and reduces operation and maintenance costs. The feedback adjustment mechanism enables the system to dynamically optimize the solution according to the actual operating status, improving the operational stability and efficiency of the power station.

[0039] Furthermore, the obtaining of N key nodes of the photovoltaic power station includes: performing three-dimensional modeling on the distribution design information to generate a three-dimensional model of the photovoltaic power station; dividing the three-dimensional model of the photovoltaic power station according to a preset grid size to obtain multiple initial photovoltaic power station grid node sets; performing maintenance criticality analysis on the multiple initial photovoltaic power station grid node sets respectively to determine multiple node maintenance criticality coefficients; performing key node screening on the multiple initial photovoltaic power station grid node sets based on the multiple node maintenance criticality coefficients to obtain the N key nodes of the photovoltaic power station.

[0040] Specifically, the preset grid size refers to the size of the segmentation unit selected when dividing the three-dimensional model into grids. These grids are used to perform zoning analysis on the power station, and each grid node contains information about equipment or components within a certain range. The initial PV power station grid node set refers to the set of grid nodes after division, which contains information about each sub-area of ​​the PV power station for subsequent analysis and screening. The node maintenance criticality coefficient is a numerical indicator used to quantify the importance of each grid node. The higher the value, the greater the impact of the node on the operation of the PV power station.

[0041] First, collect the distribution design information of the target PV power station, such as the arrangement of PV modules, the connection layout of the inverter and the distribution line, and the terrain conditions. Obtain this information through CAD design drawings, GIS geographic information files, or on-site surveying and mapping data. Then, use 3D modeling tools (such as AutoCAD, SketchUp, Revit, or MATLAB 3D toolkit) to convert this information into a 3D model of the PV power station. The 3D model can fully display the spatial distribution of PV modules and the surrounding environment, which is convenient for subsequent analysis.

[0042] Then, the 3D model is meshed. Meshing can be done based on a preset grid size, using professional tools (such as ArcGIS, MATLAB, or Open3D library in Python) to divide the PV power plant model into multiple sub-areas. Each grid node contains component or equipment information in the sub-area, such as the number of components, inverter location, transmission line status, etc. Assuming that the PV power plant covers an area of ​​1,000 square meters and the grid size is set to 10 meters × 10 meters, it will be divided into 100 grid nodes.

[0043] Next, the maintenance criticality analysis of these initial grid node sets is performed. The analysis parameters include the importance of the equipment in the node, the operating load, the historical failure rate, etc., and data analysis tools (such as Python's pandas library and R language statistical tools) are used for calculation.

[0044] Finally, according to the maintenance criticality coefficient, set the threshold to filter out N key nodes. For example, the screening rule is that the nodes with a criticality coefficient greater than 0.8 are taken as the priority maintenance targets. Sorting and filtering can be achieved using MATLAB or Python, and the output result is N key nodes of the photovoltaic power station, such as inverter connection points, main transmission line points, etc.

[0045] By performing three-dimensional modeling, meshing and key node screening of photovoltaic power stations, this method effectively solves the problems of inaccurate node identification and high operation and maintenance costs in traditional power station maintenance.

[0046] Furthermore, the obtaining of N networking node control modules includes: obtaining control module attribute information, the control module attribute information including data perception requirements, data processing requirements and data transmission requirements; performing control demand analysis on the N key nodes of the photovoltaic power station according to the control module attribute information, and determining N node control demand attribute parameters; performing control module configuration analysis based on the N node control demand attribute parameters, and obtaining N node control module configuration parameters; and performing control module deployment on the N key nodes of the photovoltaic power station in turn based on the N node control module configuration parameters, and obtaining the N networking node control modules.

[0047] Specifically, the control module attribute information refers to the functional requirements required by the control module when performing monitoring and maintenance tasks, mainly including: Data perception requirements: the scope and type of data collection, such as temperature, current, voltage, radiation intensity, etc. Data processing requirements: data analysis and computing power requirements, such as whether it is necessary to support fault diagnosis, real-time data processing or complex algorithm calculations. Data transmission requirements: data transmission mode and bandwidth requirements, such as whether high-frequency, long-distance or large data volume transmission capabilities are required. Among them, control demand analysis refers to the refinement and matching analysis of the functional requirements of the key nodes of the photovoltaic power station to determine the specific control module functions required for each node. Node control demand attribute parameters refer to the specific parameters of the control requirements of each key node, including perception accuracy, processing power, transmission rate, etc., which are used to guide the configuration of the module.

[0048] First, the attribute information of the control module is obtained, which defines the control module's capability requirements in data perception, processing, and transmission. For example, a control module needs to perceive ambient temperature and voltage fluctuations, process complex fault diagnosis algorithms, and support wireless data transmission. These requirements usually come from the operating scenarios and target functions of the photovoltaic power station, and can be determined through operation and maintenance historical data analysis or simulation testing.

[0049] Next, for each key node, control demand analysis is performed according to the attribute information of the control module to determine the specific demand attribute parameters of the node. Control demand analysis can be performed in combination with the device characteristics and operating conditions of the node. For example, for high-load inverter nodes, it is necessary to monitor multiple parameters such as voltage and current, and have strong fault analysis capabilities; for ordinary photovoltaic module nodes, it is only necessary to monitor temperature or radiation intensity, and the processing capability requirements are relatively low. Through data analysis tools (such as the pandas library in Python and MATLAB analysis tools), the operating parameters and historical data of each node are counted to extract the demand attribute parameters.

[0050] Then, according to these required parameters, the configuration analysis of the control module is performed. The configuration analysis needs to match the appropriate hardware and software resources. For example: select high-precision current sensors, embedded computing chips, and communication modules that support high-speed data transmission for high-load nodes; configure basic sensors and low-power communication modules for low-load nodes. Hardware design tools (such as NI LabVIEW, Altium Designer) and software configuration tools can be used for simulation and testing.

[0051] Finally, according to the configuration plan, control modules are deployed in sequence at key nodes of the photovoltaic power station. Each module performs tasks according to its configuration parameters to form a complete set of network node control modules. During deployment, it is necessary to ensure smooth communication between modules. Commonly used networking protocols include ZigBee, LoRa, Wi-Fi, etc.

[0052] Through this demand analysis and configuration deployment method, the deployment efficiency and operating performance of the photovoltaic power station control module can be significantly improved. The control module of each node is accurately configured according to its needs, avoiding the problem of resource waste and insufficient performance. At the same time, the collaboration of networking nodes enhances the distributed maintenance capability of the system, shortens the fault detection response time, and improves the operating reliability and maintenance efficiency of the photovoltaic power station.

[0053] Furthermore, the method of obtaining N key node fault analysis parameters includes: obtaining an abnormal data identifier and a data preprocessing program through the data processing unit; performing abnormal data identification on the N key node working data streams based on the abnormal data identifier to obtain abnormal node working data, wherein the abnormal node working data includes inconsistent data, missing data and data wild values; performing data preprocessing on the abnormal node working data based on the data preprocessing program to obtain N standard key node working data streams; performing distributed fault analysis on the N standard key node working data streams to obtain N key node fault analysis parameters.

[0054] Specifically, the abnormal data identifier is used to detect abnormal data that does not conform to normal values ​​in the working data stream. Abnormal data includes inconsistent data, missing data and data wild values. Specifically, the data preprocessor is used to correct or process the identified abnormal data to make it meet the analysis requirements. The processing methods include filling missing values, smoothing data wild values, and normalizing data formats. The standard key node working data stream is a data stream processed by the data preprocessor. The data has been cleaned and corrected, and has consistency and integrity for subsequent fault analysis.

[0055] First, the abnormal data identifier and data preprocessing program are loaded through the data processing unit. Specifically, the Z-score detection algorithm based on statistical methods is used to identify outliers that exceed the reasonable range; the rule-based logical checking algorithm is used to find inconsistencies in the data; and the time series-based model is used to identify patterns of missing data.

[0056] Next, the abnormal data identifier is used to check the working data streams of N key nodes one by one to identify the anomalies in the data. For example, the current data stream of a photovoltaic module normally fluctuates between 5A and 15A, but the data at a certain time point suddenly shows 50A, which is obviously a data wild value. In addition, if some time points in the data stream are completely null, it is missing data; and when the relationship between voltage and current does not conform to physical laws, it is inconsistent data.

[0057] Subsequently, the data preprocessing program is used to repair the abnormal data. For missing data, interpolation methods (such as linear interpolation and Lagrange interpolation) can be used to fill in; for data outliers, pruning methods can be used to adjust them to the normal range; for inconsistent data, logical rules can be used to correct or eliminate them. For example, if the temperature data of a node is missing, it can be filled with the average value of the previous and next time points; if a voltage data value is negative, the value can be eliminated and marked as an invalid point.

[0058] After data preprocessing, N standard key node working data streams are generated. These data streams have consistency and integrity and can be input into the distributed fault analysis module. Each data processing unit independently runs a distributed analysis algorithm. Common methods include trend analysis based on time series, machine learning models (such as support vector machines SVM, neural networks) for anomaly detection, or rule-based fault diagnosis models. The goal of the analysis is to identify node operation anomalies, such as unstable power output, current overload, or abnormal equipment temperature. Ultimately, the results of the distributed analysis will be output as fault analysis parameters for N key nodes.

[0059] Through this step, the quality of the working data stream of the photovoltaic power station has been significantly improved, the interference of abnormal data on fault analysis has been eliminated, and the accuracy and reliability of the analysis results have been improved. In addition, this method supports real-time processing of large-scale data streams, which helps photovoltaic power stations achieve efficient and intelligent operation and maintenance.

[0060] Furthermore, obtaining N key node fault analysis parameters includes: performing fault task analysis on the N key nodes of the photovoltaic power station to obtain N key node fault analysis task sets; acquiring the photovoltaic power station fault analysis task model library through the data processing unit; matching the N key node fault analysis task sets with the photovoltaic power station fault analysis task model library respectively to obtain N key node fault analysis model sets; performing distributed fault analysis on the N standard key node working data streams based on the N key node fault analysis model sets to obtain the N key node fault analysis parameters.

[0061] Specifically, the fault analysis task set refers to the output results of fault task analysis, which represents the list of faults occurring at each key node. Each task describes a fault type and its characteristics, such as overload detection task, voltage fluctuation detection task, etc. The fault analysis task model library is a model library stored in the data processing unit, which contains a variety of analysis models for common faults in photovoltaic power plants, such as overload analysis model, power fluctuation model, temperature anomaly model, etc. The model library provides technical support for task matching.

[0062] First, the fault task analysis is performed on N key nodes. The data processing unit will analyze the type of fault that occurs in the node based on the device type of the node (such as photovoltaic modules, inverters, distribution boxes, etc.) and the characteristics of the standard working data flow. For example, photovoltaic modules face abnormal irradiation and overheating problems, and inverters have unstable output voltage or overload faults. After analysis, the task set of each key node contains multiple fault types.

[0063] Then, the photovoltaic power station fault analysis task model library is called from the data processing unit. This model library contains analysis models for different tasks, such as trend anomaly detection models using time series analysis, or fault classification models based on machine learning (such as random forests and deep neural networks). By matching the characteristics of task requirements and model functions, the task set of each key node is matched with the model in the model library to screen out the appropriate model set.

[0064] Next, based on the model set, a distributed fault analysis is performed on the standard key node working data stream. For example, a key node task set includes an overheating detection task, and the matching model is a temperature prediction model based on regression analysis; another node task set includes a voltage fluctuation task, and the matching model is a classification model based on a support vector machine. By inputting a standardized data stream, the analysis results are output as the fault type and parameters of each node. Finally, the analysis results are integrated to form the fault analysis parameters of N key nodes. For example:

[0065] Through fault task analysis, model matching and distributed fault analysis, the fault type and parameters of each key node of the photovoltaic power station can be accurately identified. This method not only improves the efficiency of fault diagnosis, but also reduces the misjudgment rate of traditional manual analysis. In addition, customized analysis for different nodes is achieved through model matching, which ensures the accuracy of the diagnosis results and provides reliable support for intelligent operation and maintenance.

[0066] Furthermore, the acquisition of the photovoltaic power station fault analysis task model library through the data processing unit includes: collecting and acquiring a photovoltaic power station fault data set, extracting analysis tasks from the photovoltaic power station fault data set, and obtaining a photovoltaic power station fault analysis task set; constructing a deep learning network library, matching the analysis characteristic information of the photovoltaic power station fault analysis task set with the deep learning network library respectively, and obtaining a fault analysis task matching learning network set; correlating and shunting the photovoltaic power station fault data sets respectively based on the photovoltaic power station fault analysis task set, and obtaining a fault analysis task associated data set; performing distributed training on the fault analysis task associated data set based on the fault analysis task matching learning network set, constructing a photovoltaic power station fault analysis task model library and storing it in the data processing unit.

[0067] Specifically, the photovoltaic power station fault data set refers to the data set related to the faults that occurred during the operation of the photovoltaic power station, which usually includes historical fault records, equipment operating status, environmental factors (such as temperature, radiation intensity, etc.), sensor data when the fault occurred, etc. These data provide rich background information for fault diagnosis. The deep learning network library is a collection of multiple deep learning models, each of which is trained and optimized for different tasks and data types. For example, convolutional neural networks (CNNs) can be used to process image data, and recurrent neural networks (RNNs) can be used to process time series data. The library is used to match the extracted analysis tasks with suitable deep learning models. The fault analysis task matching learning network set refers to selecting and matching the most suitable deep learning network model according to the analysis characteristics in the fault analysis task set to form a dedicated learning network for each fault task. The fault analysis task association data set refers to the data association and classification of the photovoltaic power station fault data set based on the fault task analysis, and the data is subdivided according to the fault type, cause and characteristics to generate a specific data set required for each analysis task.

[0068] First, a fault data set is collected from the historical data of the photovoltaic power station. This can be extracted through sensor records and fault logs, such as the voltage, current, temperature changes of the photovoltaic panels and the time of the fault. By analyzing the data characteristics, the main fault types and their characteristics (such as overload faults and voltage fluctuations) are identified, and the task set is generated.

[0069] Based on the characteristic information in the task set, call the pre-built deep learning network library. For example, for time series data, select the LSTM model; for abnormal classification data, select the multi-layer perceptron (MLP) model. Match the task characteristics with the models in the network library one by one to obtain the learning network set. For example, the "detect current fluctuation" task can be matched with the LSTM network, and the "identify image damage" task can be matched with the CNN network.

[0070] According to the task requirements, the fault data set is assigned to the corresponding learning network set. For example, the data related to temperature anomalies is assigned to the training set of the overheating detection model, and the voltage fluctuation data is assigned to the training set of the fluctuation detection model. The data is cleaned and diverted through Python tools (such as pandas and NumPy) to form a subset of related data for each task.

[0071] The model is trained in parallel on multiple computing nodes using a deep learning framework (such as TensorFlow, PyTorch). The learning network corresponding to each task is optimized with the support of the shunted data to generate a specialized fault analysis model. After the training is completed, the optimized learning network set is integrated to generate a fault analysis task model library. The library is stored in the data processing unit and can be called at any time for real-time operation data analysis of the photovoltaic power station.

[0072] Through this distributed training, the fault types in photovoltaic power plants can be quickly and accurately identified, reducing the cost of manual monitoring, and achieving real-time fault warning and intelligent maintenance. This improves the operational efficiency and fault response speed of photovoltaic power plants.

[0073] Furthermore, the obtaining of N key node fault maintenance plans includes: performing maintenance strategy analysis on the N key node fault analysis parameter sets through the control cloud center to obtain N node fault maintenance strategies; performing associated data search based on the N node fault maintenance strategies to construct N node fault maintenance parameter spaces; using the N key node fault analysis parameter sets as constraint parameters, performing global comparison and optimization in the N node fault maintenance parameter spaces to obtain the N key node fault maintenance plans.

[0074] Specifically, the fault maintenance parameter space refers to a multidimensional parameter set formed based on the fault maintenance strategy, which is used to describe the combination of maintenance measures and their corresponding resource requirements, time costs, risk levels, etc. For example, for inverter fault maintenance, it may include "circuit switching strategy, parts replacement strategy, on-site maintenance strategy" and corresponding cost or benefit indicators.

[0075] First, the control cloud center receives and summarizes the N key node fault analysis parameter sets, compares the fault characteristics of each node with the existing maintenance experience and technical conditions through maintenance strategy analysis, and generates N node fault maintenance strategies. Subsequently, for different fault maintenance strategies, the cloud center conducts related data search, collects historical maintenance cases, spare parts information, personnel skills and cost-effectiveness and other related information to construct N node fault maintenance parameter spaces. Next, the N key node fault analysis parameter sets are used as constraints, combined with the maintenance parameter space corresponding to each node, and a global optimization algorithm (such as genetic algorithm or particle swarm algorithm) is used for multi-dimensional comparison and optimization. The algorithm evaluates the time, cost, resource occupation, risk and other dimensions of different maintenance measures, and selects the maintenance plan with the highest comprehensive value within the constraints. Finally, N key node fault maintenance plans are output to provide an execution basis for subsequent distributed maintenance.

[0076] By analyzing the maintenance strategies, searching related data, and performing global comparison and optimization of the fault analysis parameters of N key nodes, it is possible to provide targeted maintenance plans for each node in a relatively short period of time.

[0077] In summary, the distributed maintenance method for a photovoltaic power station provided by the embodiment of the present application has the following technical effects:

[0078] 1. Through key node data collection, analysis and cloud center maintenance plan formulation, accurate operation and maintenance of photovoltaic power stations are achieved, maintenance blind spots are effectively reduced, and fault response speed and operation and maintenance efficiency are improved.

[0079] 2. Fault task analysis and model matching enable highly targeted fault analysis, greatly improving the diagnostic efficiency of various fault types in photovoltaic power plants, reducing the occurrence of false alarms and missed alarms, and providing accurate guidance for subsequent maintenance.

[0080] 3. The model library is built using deep learning network libraries and distributed training, which optimizes the performance of fault analysis models, improves the efficiency of large-scale data processing and analysis in photovoltaic power plants, and lays a solid foundation for intelligent operation and maintenance.

[0081] Embodiment 2 is based on the same inventive concept as a distributed maintenance method for a photovoltaic power station in the above embodiment. Figure 2 As shown, an embodiment of the present application provides a distributed maintenance system for a photovoltaic power station, the system comprising: a key node identification module 11, used to obtain the distribution design information of the target photovoltaic power station, perform key node identification on the distributed design information, and obtain N key nodes of the photovoltaic power station; a control module deployment module 12, used to sequentially deploy control modules on the N key nodes of the photovoltaic power station, and obtain N networking node control modules, each node control module in the N networking node control modules includes a data sensing unit, a data processing unit and a data transmission unit; a distributed fault analysis module 13, used to collect and obtain N key node working data streams through the data sensing unit, perform distributed fault analysis on the N key node working data streams based on the data processing unit, and obtain N key node fault analysis parameters; a maintenance analysis module 14, used to integrate the N key node fault analysis parameter sets into the control cloud center through the data transmission unit for maintenance analysis, and obtain N key node fault maintenance plans; an adjustment and optimization module 15, used to perform distributed maintenance and feedback adjustment and optimization on the target photovoltaic power station based on the N key node fault maintenance plans.

[0082] Furthermore, the key node identification module 11 is also used to perform the following steps: perform three-dimensional modeling of the distribution design information to generate a three-dimensional model of a photovoltaic power station; divide the three-dimensional model of the photovoltaic power station according to a preset grid size to obtain multiple initial photovoltaic power station grid node sets; perform maintenance criticality analysis on the multiple initial photovoltaic power station grid node sets respectively to determine multiple node maintenance criticality coefficients; perform key node screening on the multiple initial photovoltaic power station grid node sets based on the multiple node maintenance criticality coefficients to obtain the N photovoltaic power station key nodes.

[0083] Furthermore, the control module deployment module 12 is also used to perform the following steps: obtain control module attribute information, the control module attribute information including data perception requirements, data processing requirements and data transmission requirements; perform control demand analysis on the N key nodes of the photovoltaic power station according to the control module attribute information, and determine the N node control demand attribute parameters; perform control module configuration analysis based on the N node control demand attribute parameters to obtain N node control module configuration parameters; and perform control module deployment on the N key nodes of the photovoltaic power station in turn based on the N node control module configuration parameters to obtain the N networking node control modules.

[0084] Furthermore, the distributed fault analysis module 13 is also used to perform the following steps: obtaining an abnormal data identifier and a data preprocessing program through the data processing unit; performing abnormal data identification on the N key node working data streams based on the abnormal data identifier to obtain abnormal node working data, and the abnormal node working data includes inconsistent data, missing data and data wild values; performing data preprocessing on the abnormal node working data based on the data preprocessing program to obtain N standard key node working data streams; performing distributed fault analysis on the N standard key node working data streams to obtain N key node fault analysis parameters.

[0085] Furthermore, the distributed fault analysis module 13 is also used to perform the following steps: perform fault task analysis on the N key nodes of the photovoltaic power station to obtain N key node fault analysis task sets; obtain the photovoltaic power station fault analysis task model library through the data processing unit; match the N key node fault analysis task sets with the photovoltaic power station fault analysis task model library respectively to obtain N key node fault analysis model sets; perform distributed fault analysis on the N standard key node working data streams based on the N key node fault analysis model sets to obtain the N key node fault analysis parameters.

[0086] Furthermore, the distributed fault analysis module 13 is also used to perform the following steps: collect and obtain a photovoltaic power station fault data set, extract analysis tasks for the photovoltaic power station fault data set, and obtain a photovoltaic power station fault analysis task set; construct a deep learning network library, and match the analysis characteristic information of the photovoltaic power station fault analysis task set with the deep learning network library to obtain a fault analysis task matching learning network set; associate and shunt the photovoltaic power station fault data sets based on the photovoltaic power station fault analysis task set to obtain a fault analysis task associated data set; perform distributed training on the fault analysis task associated data set based on the fault analysis task matching learning network set, construct a photovoltaic power station fault analysis task model library and store it in the data processing unit.

[0087] Furthermore, the maintenance analysis module 14 is also used to perform the following steps: performing maintenance strategy analysis on the N key node fault analysis parameter sets through the control cloud center to obtain N node fault maintenance strategies, performing related data search based on the N node fault maintenance strategies, and constructing N node fault maintenance parameter spaces; using the N key node fault analysis parameter sets as constraint parameters, performing global comparison and optimization in the N node fault maintenance parameter spaces, and obtaining the N key node fault maintenance plans.

[0088] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0089] Furthermore, the first or second mentioned above not only represents an order relationship, but also represents a specific concept, and / or refers to the selection of multiple elements individually or in whole. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application intends to include these modifications and variations.

Claims

1. A distributed maintenance method for a photovoltaic power station, characterized in that: The method comprises: Obtaining distribution design information of a target photovoltaic power station, identifying key nodes of the distribution design information, and obtaining N key nodes of the photovoltaic power station; Control modules are deployed sequentially on the N key nodes of the photovoltaic power station to obtain N networking node control modules, wherein each node control module in the N networking node control modules includes a data sensing unit, a data processing unit and a data transmission unit; Acquire N key node working data streams through the data sensing unit, perform distributed fault analysis on the N key node working data streams based on the data processing unit, and obtain N key node fault analysis parameters; Integrating the N key node fault analysis parameter sets into the control cloud center through the data transmission unit for maintenance analysis to obtain N key node fault maintenance solutions; Distributed maintenance and feedback adjustment optimization are performed on the target photovoltaic power station based on the N key node failure maintenance solutions.

2. A distributed maintenance method for a photovoltaic power station as claimed in claim 1, characterized in that: The obtaining of N key nodes of a photovoltaic power station includes: Performing three-dimensional modeling on the distribution design information to generate a three-dimensional model of a photovoltaic power station; Dividing the photovoltaic power station three-dimensional model according to a preset grid size to obtain a plurality of initial photovoltaic power station grid node sets; Performing maintenance criticality analysis on the multiple initial photovoltaic power station grid node sets respectively to determine multiple node maintenance criticality coefficients; Based on the multiple node maintenance criticality coefficients, key node screening is performed on the multiple initial photovoltaic power station grid node sets to obtain the N photovoltaic power station key nodes.

3. A distributed maintenance method for a photovoltaic power station as claimed in claim 1, characterized in that: The obtaining of N networking node control modules includes: Acquire control module attribute information, wherein the control module attribute information includes data perception requirements, data processing requirements, and data transmission requirements; According to the control module attribute information, control demand analysis is performed on the N key nodes of the photovoltaic power station respectively to determine the control demand attribute parameters of the N nodes; Performing control module configuration analysis based on the N node control requirement attribute parameters to obtain N node control module configuration parameters; Based on the N node control module configuration parameters, control modules are deployed in sequence on the N key nodes of the photovoltaic power station to obtain the N networking node control modules.

4. A distributed maintenance method for a photovoltaic power station as claimed in claim 1, characterized in that: The N key node fault analysis parameters are obtained, including: Acquiring an abnormal data identifier and a data preprocessing program through the data processing unit; Based on the abnormal data identifier, the N key node working data streams are subjected to abnormal data identification to obtain abnormal node working data, wherein the abnormal node working data includes inconsistent data, missing data and data outliers; Based on the data preprocessing program, the abnormal node working data is preprocessed to obtain N standard key node working data streams; Distributed fault analysis is performed on the N standard key node working data streams to obtain N key node fault analysis parameters.

5. A distributed maintenance method for a photovoltaic power station as claimed in claim 4, characterized in that: The N key node fault analysis parameters are obtained, including: Performing fault task analysis on the N key nodes of the photovoltaic power station to obtain a fault analysis task set of N key nodes; Acquiring a photovoltaic power station fault analysis task model library through the data processing unit; Based on the N key node fault analysis task sets, the N key node fault analysis task sets are matched with the photovoltaic power station fault analysis task model library to obtain N key node fault analysis model sets; Based on the N key node fault analysis model sets, distributed fault analysis is performed on the N standard key node working data streams to obtain the N key node fault analysis parameters.

6. A distributed maintenance method for a photovoltaic power station as claimed in claim 5, characterized in that: The step of obtaining a photovoltaic power station fault analysis task model library through the data processing unit includes: Collecting and acquiring a photovoltaic power station fault data set, extracting analysis tasks from the photovoltaic power station fault data set, and obtaining a photovoltaic power station fault analysis task set; Constructing a deep learning network library, matching the analysis characteristic information of the photovoltaic power station fault analysis task set with the deep learning network library respectively, and obtaining a fault analysis task matching learning network set; Based on the photovoltaic power station fault analysis task set, the photovoltaic power station fault data set is respectively associated and split to obtain a fault analysis task associated data set; Based on the fault analysis task matching learning network set, the fault analysis task associated data set is subjected to distributed training, and a photovoltaic power station fault analysis task model library is constructed and stored in the data processing unit.

7. A distributed maintenance method for a photovoltaic power station as claimed in claim 1, characterized in that: The obtaining of N key node fault maintenance solutions includes: Performing maintenance strategy analysis on the N key node fault analysis parameter sets through the control cloud center to obtain N node fault maintenance strategies; Performing a related data search based on the N node fault maintenance strategies to construct an N node fault maintenance parameter space; The N key node fault analysis parameter sets are used as constraint parameters, and global comparison and optimization are performed in the N node fault maintenance parameter spaces to obtain the N key node fault maintenance solutions.

8. A distributed maintenance system for a photovoltaic power station, characterized in that: A system for executing a distributed maintenance method for a photovoltaic power station according to any one of claims 1 to 7, comprising: A key node identification module is used to obtain the distribution design information of the target photovoltaic power station, identify the key nodes of the distribution design information, and obtain N key nodes of the photovoltaic power station; A control module deployment module, used to sequentially deploy control modules on the N key nodes of the photovoltaic power station to obtain N networking node control modules, each node control module in the N networking node control modules includes a data sensing unit, a data processing unit and a data transmission unit; A distributed fault analysis module, used to collect and acquire N key node working data streams through the data perception unit, and perform distributed fault analysis on the N key node working data streams based on the data processing unit to obtain N key node fault analysis parameters; A maintenance analysis module, used to integrate the N key node fault analysis parameter sets into the control cloud center through the data transmission unit for maintenance analysis, and obtain N key node fault maintenance solutions; The adjustment and optimization module is used to perform distributed maintenance and feedback adjustment and optimization on the target photovoltaic power station based on the N key node fault maintenance solutions.

Images