A smart operation and maintenance control system for photovoltaic power plants

Abstract

The present invention discloses an intelligent operation and maintenance control system applied to a photovoltaic field, which relates to the field of photovoltaic power station operation and maintenance, including a real-time acquisition module, a photovoltaic cloud platform, a fault maintenance module, an adjustment control module and a multi-device management module. The real-time acquisition module acquires the operating status and environmental parameters of photovoltaic field equipment in real time, and performs preliminary edge processing and analysis on the acquired data. The photovoltaic cloud platform performs distributed fault monitoring, diagnosis and early warning on photovoltaic field equipment by constructing a lightweight twin fault supervision model. The fault maintenance module makes fault maintenance decisions. The adjustment control module performs intelligent control and regulation on the operation of photovoltaic field equipment. The multi-device management module performs visual monitoring on photovoltaic field equipment. The problem solved is that equipment data is isolated, which leads to difficulty in locating abnormal components, low operation and maintenance efficiency and great difficulty.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power station operation and maintenance, and in particular to an intelligent operation and maintenance control system applied to a photovoltaic field. Background Art

[0002] Photovoltaic power stations are power generation systems based on solar energy using materials such as crystalline silicon panels, and are an important part of my country's power resource production system. In recent years, photovoltaic power generation, as one of the important representatives of clean energy, has received more and more attention and recognition. With the rapid development of the photovoltaic industry, the scale of photovoltaic sites has continued to expand, and the number of power stations has also continued to increase. The construction area of ​​the sites is wide, mostly built in remote areas, with complex geographical conditions and long equipment spacing, which also puts higher requirements on the operation and maintenance of power stations. The operation and maintenance management of photovoltaic power stations mainly relies on manual work, which increases the fault response time, long manual work time and high intensity, and increases the maintenance volume and maintenance cost. Therefore, an intelligent operation and maintenance control system applied to photovoltaic sites has emerged to replace the traditional manual operation and maintenance methods and improve the operation and maintenance efficiency.

[0003] At present, photovoltaic power station operation and maintenance technologies mainly used in photovoltaic fields are based on drone inspections and infrared image monitoring. The abnormal parts of photovoltaic power stations are located through image technology, while other factors of photovoltaic power stations, such as current and voltage, are ignored. In addition, photovoltaic power stations include solar panels, inverters, battery energy storage systems, etc. There are many independent devices, and the data cannot be associated. Therefore, the existing technology is difficult to quickly grasp the abnormal photovoltaic components during the operation and maintenance process, and it is difficult to coordinate the operation of the entire photovoltaic field. The operation and maintenance efficiency is low and the difficulty is great.

[0004] Therefore, an intelligent operation and maintenance control system applied to photovoltaic fields is needed to solve the above problems. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention discloses an intelligent operation and maintenance control system applied to photovoltaic fields; it can realize comprehensive monitoring, diagnosis and early warning of photovoltaic field equipment, and has remote control and intelligent adjustment functions, which helps to improve the efficiency and stability of photovoltaic power generation systems, and has a high degree of automation and intelligence.

[0006] The present invention adopts the following technical solutions:

[0007] An intelligent operation and maintenance control system applied to a photovoltaic field, the intelligent operation and maintenance control system comprising:

[0008] A real-time acquisition module, which collects the operating status and environmental parameters of photovoltaic field equipment in real time through a sensor network, and performs preliminary processing and analysis on the collected data by deploying edge computing nodes. The photovoltaic field equipment includes at least solar panels, inverters, and battery energy storage system equipment. The operating status and environmental parameters include at least current, voltage, temperature, power, and light intensity.

[0009] Photovoltaic cloud platform, which receives and stores edge processing data in real time through a full-station accelerated communication network, and performs distributed fault monitoring, diagnosis and early warning on the photovoltaic field equipment by building a lightweight twin fault supervision model;

[0010] A fault maintenance module, which makes fault maintenance decisions through a genetic retrieval enhancement model, which adaptively generates a maintenance plan based on fault monitoring, diagnosis and early warning information, and performs fault maintenance control through remote control;

[0011] An adjustment control module, wherein the adjustment control module intelligently controls and adjusts the operation of the photovoltaic field equipment through a fuzzy PID controller, and the fuzzy PID controller performs switch control, load control, power regulation control and battery energy storage management of the photovoltaic field equipment according to real-time operation data of the photovoltaic field, historical fault maintenance data and external environmental parameters;

[0012] A multi-device management module, which realizes visual monitoring of the photovoltaic field equipment through a visual integrated management platform, and realizes centralized control and management of multiple photovoltaic field equipment nodes through a cloud server;

[0013] The output end of the real-time acquisition module is connected to the input end of the photovoltaic cloud platform, the output end of the photovoltaic cloud platform is connected to the input end of the fault maintenance module, the output ends of the real-time acquisition module and the fault maintenance module are connected to the input end of the adjustment control module, the output end of the real-time acquisition module is connected to the input end of the multi-device management module, and the output end of the fault maintenance module is connected to the input end of the multi-device management module.

[0014] Furthermore, the photovoltaic cloud platform builds a lightweight fault supervision model based on the physical information and dynamic sensor information of the photovoltaic field. The lightweight fault supervision model includes a physical entity layer, a virtual entity layer, a data interaction layer, a twin model layer, a model lightweight layer and an application layer. The physical entity layer drives the virtual entity to run in real time through real-time collected data. The virtual entity layer simulates the photovoltaic field subsystem, equipment component structure and dynamic operation status under different working conditions based on historical collected data and real-time collected data. The data interaction layer is used to transfer the data of the physical entity layer and the virtual entity layer to the twin model layer for processing. The twin model layer uses machine learning to The actual data is feature matched and compared with the simulated data to realize fault monitoring, diagnosis and prediction of the equipment operation status. The model lightweight layer performs lightweight optimization on the lightweight fault supervision model through model pruning and parameter optimization. The application layer is used to summarize, integrate and interpret the output of the lightweight fault supervision model, and push alarm information and early warning information to management personnel or maintenance personnel. The twin model layer includes a fault detection module, a fault diagnosis module and a fault prediction module. The output end of the fault detection module is connected to the input end of the fault diagnosis module, and the output end of the fault diagnosis module is connected to the input end of the fault prediction module.

[0015] Furthermore, the fault detection module dynamically detects the power generation efficiency of the photovoltaic area based on the actual collected data and the simulated collected data, and detects the fault of the photovoltaic area by comparing the actual power generation efficiency of the photovoltaic area with the simulated power generation efficiency. The calculation formula of the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic area is:

[0016] (1)

[0017] In formula (1), represents the actual power generation efficiency of the photovoltaic field at the i-th sampling moment, r represents the actual photovoltaic field, represents the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, m represents the simulated photovoltaic field, Indicates the ith sampling instantaneous value of the actual output power of the photovoltaic module, , Represents the ith sampling instantaneous value of the simulated output power of the photovoltaic module, represents the sampling period, Indicates the nominal power of the photovoltaic module. Indicates the instantaneous value of solar irradiance sampling on the front of the photovoltaic module. Indicates the solar irradiance under standard test conditions, , Represents the linear sampling parameter, which is used to adjust the deviation between the simulated output power and the actual output power of the photovoltaic module; the difference formula between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field is:

[0018] (2)

[0019] In formula (2), Indicates the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, Indicates the maximum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the maximum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, It is used to eliminate weight deviations caused by different data magnitudes and numerical ranges. When the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the sampling moment is greater than the similarity threshold, it is determined that the actual power generation efficiency of the photovoltaic field at the sampling moment is abnormal.

[0020] Furthermore, the fault diagnosis module classifies and calculates the illumination, temperature, output voltage and current of the photovoltaic device and the working state of the inverter based on the energy efficiency differences between the photovoltaic components. First, the actual output power of the photovoltaic array is obtained according to the output voltage and current of the photovoltaic array. At the same time, the rated output power of the photovoltaic array in the current environment is calculated according to the light intensity, ambient temperature and photovoltaic array temperature, and the actual output power of the photovoltaic array is compared with the rated output power to obtain the conversion efficiency of the photovoltaic array. Then, the conversion efficiency of the inverter is calculated according to the input power and output power of the inverter, and the conversion efficiency of the inverter is monitored and diagnosed to determine whether the capacitor and the switch tube are faulty.

[0021] Furthermore, the fault prediction module collects power timing curve data of the photovoltaic field equipment to extract the power generation efficiency and power output of the equipment in different time periods, and performs feature analysis on the power timing curve through time series analysis to discover potential inefficient power generation sites and issue early warnings.

[0022] Furthermore, the genetic retrieval enhancement model adaptively generates a maintenance plan based on a large model retrieval enhancement tool and a genetic algorithm, and the working method of the genetic retrieval enhancement model includes the following steps:

[0023] S1. Data input and initialization. The large model retrieval enhancement tool constructs a fault maintenance knowledge graph by retrieving the fault maintenance history records of the photovoltaic field and the expert knowledge base, and classifies the fault maintenance knowledge graph according to the equipment fault type and maintenance method, and preliminarily matches the fault monitoring, diagnosis and warning information with the fault maintenance knowledge graph to generate an initial fault maintenance strategy population;

[0024] S2, population evolution, evaluates and ranks the solutions in the initial population through the fitness function, and evolves the initial population through population individual selection, crossover and mutation operations;

[0025] S3, dynamically updating parameters, the large model retrieval enhancement tool dynamically updates the fitness function, selection strategy, crossover and mutation operations of the genetic algorithm according to the real-time collected data and the retrieval maintenance decision;

[0026] S4, iteration and convergence judgment, judging whether the maximum number of iterations or fitness value convergence is reached, if the maximum number of iterations or fitness value convergence is reached, then execute S5 operation, otherwise execute S2 operation;

[0027] S5. Output the optimal maintenance decision plan, and feed back the data and decision results obtained during the execution process to the fault maintenance knowledge graph for updating.

[0028] Furthermore, the visual integrated management platform obtains the operating parameters of the photovoltaic field equipment in real time through the high-speed serial expansion bus PCIe, and uses interactive charts, heat maps, maps and dashboards to perform real-time visual monitoring of equipment data change trends, change patterns and equipment relationships. The visual integrated management platform uses a Token user identity authentication mechanism to verify the identity of the accessing user. After successful identity authentication, the user can remotely adjust the operating parameters of the photovoltaic field equipment by inputting event functions.

[0029] Furthermore, the whole-station acceleration communication network listens to the collected data forwarding method defined by the routing type through the three-layer TCP / UDP transmission protocol and the seven-layer HTTP / HTTPS cache protocol, and distributes the data traffic to the back-end server in real time to achieve end-to-end transmission. The whole-station acceleration communication network uses the low-voltage modulation communication standard LVM-hscs to perform differential Manchester encoding on the collected data to achieve data anti-interference communication, and uses the Secure Sockets Layer protocol SSL acceleration card to encrypt and authenticate data transmission.

[0030] The beneficial effects of the present invention are:

[0031] 1. The present invention collects the operating status and environmental parameters of photovoltaic field equipment in real time, performs preliminary processing and analysis on the collected data, realizes data integration and association, and improves operation and maintenance efficiency and accuracy. At the same time, by constructing a lightweight twin fault supervision model, distributed fault monitoring, diagnosis and early warning are performed on photovoltaic field equipment, improving the accuracy of fault monitoring and diagnosis, and enhancing the reliability and stability of the system.

[0032] 2. The present invention uses a genetic retrieval enhancement model to make fault maintenance decisions, adaptively generates maintenance plans based on fault monitoring, diagnosis and early warning information, and performs fault maintenance control through remote control, thereby improving fault handling efficiency and accuracy and reducing maintenance costs and risks.

[0033] 3. The present invention realizes centralized automatic management and optimized scheduling of photovoltaic electric field equipment through a visual integrated management platform, thereby improving the operating efficiency and management level of photovoltaic power stations. At the same time, the operation of photovoltaic field equipment is intelligently controlled and adjusted through a fuzzy PID controller, and photovoltaic field equipment is controlled and managed according to real-time operation data, historical fault maintenance data and external environmental parameters of the photovoltaic field, thereby improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the overall system architecture of the present invention;

[0035] Figure 2 Schematic diagram of the architecture of the lightweight twin fault supervision model in the present invention;

[0036] Figure 3 This is a schematic diagram of the working process of the fault prediction module in the present invention;

[0037] Figure 4 Schematic diagram of the workflow of the genetic retrieval enhancement model in the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the attached embodiment of the present invention Figure 1 To Attachment Figure 4 , the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The embodiment of the present invention discloses an intelligent operation and maintenance control system applied to a photovoltaic field, wherein the intelligent operation and maintenance control system comprises: a real-time acquisition module, a photovoltaic cloud platform, a fault maintenance module, an adjustment control module and a multi-device management module;

[0040] The real-time acquisition module can collect the operating status and environmental parameters of the solar panels, inverters and battery energy storage systems of the photovoltaic power station in real time through the sensor network, such as current, voltage, temperature, power and light intensity, etc., and perform preliminary processing and analysis on the collected data by deploying edge computing nodes, and send the processing results to the photovoltaic cloud platform.

[0041] The photovoltaic cloud platform can receive and store edge processing data in real time through the full-station accelerated communication network, and build a lightweight twin fault supervision model to perform distributed fault monitoring, diagnosis and early warning for solar panels, inverters and battery energy storage systems. For example, when the power generation of a solar panel drops, the photovoltaic cloud platform can promptly diagnose the fault of the solar panel and notify the fault maintenance module to repair it through the early warning function.

[0042] The fault maintenance module can make fault maintenance decisions through the genetic retrieval enhancement model, adaptively generate maintenance plans based on fault monitoring, diagnosis and early warning information, and perform fault maintenance control through remote control. For example, when a solar panel fails, the fault maintenance module can generate a maintenance plan and disconnect the solar panel from the grid through remote control for maintenance operations.

[0043] The adjustment control module intelligently controls and adjusts the operation of the solar photovoltaic power station through a fuzzy PID controller. For example, when the light intensity changes, the adjustment control module can perform switch control, load control, power regulation control and battery energy storage management of the photovoltaic field equipment according to real-time operation data, historical fault maintenance data and external environmental parameters.

[0044] The multi-device management module can realize visual monitoring of solar photovoltaic power station equipment through a visual integrated management platform, and realize centralized control and management of multiple solar photovoltaic power station equipment nodes through a cloud server. For example, the operation status, historical fault records, and fault warning information of the photovoltaic power station can be intuitively viewed through the visual interface.

[0045] In short, this intelligent operation and maintenance control system can improve the operation efficiency and management level of solar photovoltaic power stations, reduce maintenance costs and risks, and improve system stability and reliability.

[0046] The photovoltaic cloud platform builds a lightweight fault supervision model based on the physical information and dynamic sensor information of the photovoltaic field. The lightweight fault supervision model includes a physical entity layer, a virtual entity layer, a data interaction layer, a twin model layer, a model lightweight layer and an application layer.

[0047] The physical entity layer uses a sensor network to collect real-time operating status data and environmental parameters of each subsystem and component equipment in the photovoltaic field, such as current, voltage, temperature, power, etc., and transmits the collected data to the virtual entity layer in real time for subsequent simulation and computational analysis.

[0048] The virtual entity layer simulates and evaluates the dynamic operating status of the PV field subsystems, equipment component structures and different operating conditions based on the analysis of historical and real-time collected data, so as to achieve accurate simulation and monitoring of the actual operating conditions, and further perform data analysis and processing on the output of the virtual entity layer.

[0049] The data interaction layer is used to integrate the collected data from the physical entity layer and the virtual entity layer, and pass it to the twin model layer for related processing and data calculation, and ensure data interoperability and data consistency between the physical entity layer and the virtual entity layer.

[0050] The twin model layer is mainly used to monitor, diagnose and predict faults of photovoltaic power station equipment, including fault detection module, fault diagnosis module and fault prediction module. Among them, the fault detection module is mainly used to analyze the data of the virtual entity layer and the physical entity layer, identify possible fault phenomena, and pass the detection results to the fault diagnosis module. The fault diagnosis module accurately diagnoses and locates faults by combining multi-source data and knowledge base. The fault prediction module can use machine learning algorithms to predict the future possibility of equipment failures in order to prevent and reduce unnecessary downtime.

[0051] The model lightweight layer mainly performs lightweight optimization on the extracted twin model through model pruning and parameter optimization to improve the calculation speed and efficiency of the model.

[0052] The application layer mainly interprets, analyzes and processes the output results, and feeds back the results to management personnel or maintenance personnel, as well as pushes alarm information and early warning information so that correct operation and maintenance measures and preventive measures can be taken in a timely manner.

[0053] The output end of the physical entity layer is connected to the input end of the virtual entity layer, the output ends of the physical entity layer and the virtual entity layer are connected to the input end of the data interaction layer, the output end of the data interaction layer is connected to the input end of the twin model layer, the output end of the model lightweight layer is connected to the input end of the twin model layer, and the output end of the twin model layer is connected to the input end of the application layer.

[0054] In the above embodiment, the fault detection module dynamically detects the power generation efficiency of the photovoltaic field based on the actual collected data and the simulated collected data, and detects the fault of the photovoltaic field by comparing the actual power generation efficiency of the photovoltaic field with the simulated power generation efficiency. The calculation formula of the actual power generation efficiency of the photovoltaic field and the simulated power generation efficiency is:

[0055] (1)

[0056] In formula (1), represents the actual power generation efficiency of the photovoltaic field at the i-th sampling moment, r represents the actual photovoltaic field, represents the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, m represents the simulated photovoltaic field, Indicates the ith sampling instantaneous value of the actual output power of the photovoltaic module, , Represents the ith sampling instantaneous value of the simulated output power of the photovoltaic module, represents the sampling period, Indicates the nominal power of the photovoltaic module. Indicates the instantaneous value of solar irradiance sampling on the front of the photovoltaic module. Indicates the solar irradiance under standard test conditions, , Represents the linear sampling parameter, which is used to adjust the deviation between the simulated output power and the actual output power of the photovoltaic module; the difference formula between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field is:

[0057] (2)

[0058] In formula (2), Indicates the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, Indicates the maximum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the maximum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, It is used to eliminate weight deviations caused by different data magnitudes and numerical ranges. When the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the sampling moment is greater than the similarity threshold, it is determined that the actual power generation efficiency of the photovoltaic field at the sampling moment is abnormal.

[0059] Formula related parameters: Sampling period: set according to actual needs and system configuration, for example, sampling every 5 minutes. Nominal power of PV modules: determined according to the specifications of PV modules, such as 250W, 300W, etc. Linear sampling parameters: used to adjust the deviation between the simulated output power and the actual output power of PV modules, set according to historical data and system characteristics. Similarity threshold: set according to the actual situation of the PV field and operation and maintenance requirements, used to determine whether the actual power generation efficiency is abnormal.

[0060] In a specific embodiment, the fault detection module needs the following key hardware support to run in a server environment:

[0061] High-performance server: It needs to have strong computing power and data storage capacity to handle large amounts of real-time data and complex computing tasks. It is equipped with multi-core processors and high-speed memory to ensure real-time performance and fast response. It is equipped with large-capacity hard disk or SSD storage to store historical data and run models.

[0062] Network communication hardware: The server needs to be equipped with a high-speed network interface card (NIC) to support high-speed communication with other systems (such as data acquisition systems, control systems, etc.). Network equipment such as firewalls and routers may also be required to ensure network security and reliable data transmission.

[0063] Database management system: An efficient database management system (DBMS) is run on the server to store and manage real-time data, historical data, model parameters, etc. of the photovoltaic field. The database system needs to support efficient data retrieval and query to quickly respond to fault detection needs.

[0064] Backup and recovery hardware: To ensure data integrity and reliability, the server needs to be equipped with backup storage devices (such as tape libraries, backup servers, etc.) to regularly back up data and system configurations. At the same time, a disaster recovery plan needs to be considered to ensure that services can be quickly restored in the event of a failure.

[0065] Virtualization technology: Consider using virtualization technology, such as virtual machines (VMs) or containerization technology (such as Docker), on the server to isolate different applications and services and improve resource utilization and management efficiency.

[0066] Monitoring and management software: Monitoring and management software should be installed on the server to monitor the server's operating status, performance, and network connection, and promptly identify and resolve potential problems. At the same time, a software program with a fault detection module should also be installed to achieve real-time fault detection and analysis.

[0067] Redundant power supply and cooling system: To ensure the stable operation of the server, a redundant power supply and cooling system is required to prevent downtime accidents caused by power failure or overheating.

[0068] Secure hardware: The server should be deployed in a secure environment and equipped with physical security facilities (such as access control systems, video surveillance systems, etc.) to prevent unauthorized access and damage.

[0069] For 10 photovoltaic modules in a photovoltaic field, a photovoltaic module test platform was built, and the 10 modules were connected to the circuit for testing. The test platform includes a light intensity meter, a data acquisition instrument, and voltage and current measurement equipment. The sampling period is set to 10 seconds, the nominal power is 50W, the instantaneous value of the positive solar irradiance sampling is 500W / m², the solar irradiance under standard test conditions is 1000W / m², and the linear sampling parameter is 0.95.

[0070] Some faults were artificially created, such as blocking some photovoltaic modules. According to formula (1), 10 modules were tested using sampled data. The actual power generation efficiency and simulated power generation efficiency of each module under a light intensity of 500 W / m² were calculated, and the difference was calculated and recorded in Table 1.

[0071] Table 1 Results Statistics

[0072]

[0073] The accuracy of fault judgment detected by the fault detection module is verified and compared with the accuracy of general detection methods. The comparison results are recorded in Table 2.

[0074] Table 2 Results Statistics

[0075]

[0076] The above results show that the fault detection method used in this experiment has an accuracy of 90%, while the accuracy of the general detection method is 80%. The fault detection method has a higher accuracy rate than the general detection method and can better serve the maintenance and management of photovoltaic modules.

[0077] The fault diagnosis module is based on the energy efficiency differences between photovoltaic components. Through comparative analysis of multiple parameters, it classifies and calculates the illumination, temperature, output voltage and current of the photovoltaic array, and the working status of the inverter of the photovoltaic device, thereby realizing automatic fault diagnosis and early warning of the photovoltaic power generation system.

[0078] Specifically, the fault diagnosis module first uses the output voltage and current of the photovoltaic array to calculate the actual output power of the photovoltaic array. At the same time, it combines factors such as light intensity, ambient temperature and photovoltaic array temperature to calculate the rated output power of the photovoltaic array, thereby obtaining the conversion efficiency of the photovoltaic array, further evaluating the energy efficiency differences between photovoltaic modules and discovering potential fault points.

[0079] Secondly, the fault diagnosis module monitors the input power and output power of the inverter, analyzes the conversion efficiency of the inverter, and determines whether the capacitor and switch tube are faulty based on the working status of the inverter. If there is a fault, the module will automatically warn and point out the specific location and impact range of the fault through a detailed fault report, providing a quick solution for the operation and maintenance personnel.

[0080] The fault diagnosis and early warning functions of the fault diagnosis module can improve the reliability and operation and maintenance efficiency of the photovoltaic power generation system, and reduce fault losses and equipment maintenance costs.

[0081] The implementation process of the fault prediction module includes the following steps:

[0082] Step 1. Collect power timing curve data:

[0083] The fault prediction module collects the power timing curve data of the equipment by connecting to the monitoring system of the photovoltaic power station equipment. The collected data includes information such as the power generation and efficiency of different equipment in different time periods.

[0084] Step 2. Extract equipment power generation efficiency and power:

[0085] Based on the collected data, the fault prediction module can extract the power generation efficiency and power output of different equipment in different time periods. This information can help managers understand the operating status of the equipment in a timely manner and predict equipment failures.

[0086] Step 3. Time series analysis:

[0087] By performing time series analysis on the collected power timing curves, some obvious characteristics of the equipment during operation can be found, such as power fluctuations, peak-to-valley changes, etc. By analyzing these characteristics, possible faults or problems can be found and handled in a timely manner.

[0088] Step 4. Discover potential faults:

[0089] Based on the results of time series analysis, the fault prediction module can detect potential fault phenomena, such as component aging, battery damage, pollution, etc. Through timely early warning, equipment downtime can be reduced and the efficient operation of photovoltaic power station equipment can be guaranteed.

[0090] In general, the fault prediction module can process and analyze the power timing curve data of photovoltaic power station equipment to identify possible faults or problems, and issue early warnings so that managers can take timely measures to ensure the reliability and efficiency of equipment operation.

[0091] The specific implementation steps of the working method of the genetic retrieval enhancement model are as follows:

[0092] S1. Data input and initialization: The large model retrieval enhancement tool uses the fault maintenance history records and expert knowledge base of the photovoltaic field to build a fault maintenance knowledge graph. The large model retrieval enhancement tool classifies the knowledge graph according to the equipment fault type and maintenance method, matches the fault monitoring, diagnosis and warning information with the fault maintenance knowledge graph, and generates an initial fault maintenance strategy population.

[0093] S2. Population evolution: Use the fitness function to evaluate and sort the solutions in the initial population, and perform population individual selection, crossover and mutation operations to evolve the initial population. The fitness function is a metric used in genetic algorithms to evaluate the quality of each individual (i.e., candidate solution) in the population. In the photovoltaic field fault maintenance plan, the fitness function can be defined based on multiple factors, such as maintenance cost, maintenance time, recovery efficiency, fault recurrence rate, etc. These factors can be weighted according to actual needs to form a comprehensive evaluation index.

[0094] S3. Dynamic update parameters: The large model retrieval enhancement tool dynamically updates the fitness function, selection strategy, crossover and mutation operations based on real-time collected data and retrieval maintenance decisions.

[0095] S4, Iteration and convergence judgment: judge whether the maximum number of iterations is reached or the fitness value converges. If the maximum number of iterations is reached or the fitness value converges, execute S5; otherwise execute S2. It is set that when the average fitness value of the population or the fitness value of the optimal individual changes very little (below a preset threshold) over multiple generations, the algorithm is considered to have converged. The setting of the convergence condition should take into account the actual situation of the problem and the limitations of computing resources.

[0096] S5. Output the optimal maintenance decision plan: Output the optimized optimal maintenance decision plan, and feed back the data collected during the execution and the decision effect to the fault maintenance knowledge graph for updating. Using this feedback information, the large model retrieval enhancement tool can make more accurate predictions during the next maintenance operation, thereby continuously improving the maintenance decision effect.

[0097] In general, the genetic retrieval enhancement model can be used to adaptively generate and maintain maintenance plans based on large model retrieval enhancement tools and genetic algorithms. The model continuously optimizes the maintenance plan based on the real-time collected data and feedback information to improve the efficiency and accuracy of fault maintenance. At the same time, the model can also integrate all maintenance experience and knowledge into the fault maintenance knowledge graph, so that managers and maintenance personnel can better use existing knowledge and improve the overall operation efficiency of photovoltaic power plants.

[0098] Updates to the fault maintenance knowledge graph include:

[0099] Real-time data and feedback information: Through real-time data and feedback information, we can understand the specific circumstances of the failure, the effectiveness of maintenance measures, etc. This information can be used to update the relevant nodes and relationships in the knowledge graph.

[0100] Expert knowledge: Experts can supplement and modify the knowledge graph based on new failure cases and maintenance experience.

[0101] Algorithm learning results: The genetic retrieval enhancement model can generate new and better maintenance strategies through learning and evolution, which can also be used to update the knowledge graph.

[0102] The update mechanism may include adding new nodes, modifying node attributes, adding or deleting relationships between nodes, etc.

[0103] The parameters of the genetic algorithm include selection strategy, crossover probability, mutation probability, etc. The settings of these parameters have a great impact on the performance of the algorithm. Common selection strategies include roulette wheel selection, random traversal sampling, etc. The selection strategy should ensure that excellent individuals have more opportunities to be selected for subsequent crossover and mutation operations. The crossover probability determines the probability of two individuals in the population performing a crossover operation. Too high a crossover probability may cause the algorithm to converge too early, while too low a crossover probability may cause the algorithm to converge too slowly. The mutation probability determines the probability of an individual in the population undergoing a mutation operation. An appropriate mutation probability helps maintain the diversity of the population and prevents the algorithm from falling into a local optimal solution. The specific settings of these parameters should be adjusted and optimized according to the actual situation of the problem and the limitations of computing resources.

[0104] The genetic search enhancement model needs to run in the following hardware working environments:

[0105] 1. Server: used to store data, perform calculations and processing, and provide an interface for users to perform search operations.

[0106] 2. Storage devices: used to store large amounts of data such as genomic data, reference databases, and search results.

[0107] 3. Computing equipment: used for calculation and processing, including comparison and matching of large amounts of genomic data, and analysis and screening of search results.

[0108] 4. Network equipment: used to achieve data transmission and communication, including communication with users, reference databases and other databases, etc.

[0109] 5. Database management system: used to manage and maintain reference databases and search result databases, including data backup, recovery, maintenance and optimization operations.

[0110] The above hardware working environment needs to be efficient, stable, scalable and secure to meet the needs of large-scale genomic data processing and analysis. In addition, in order to ensure the performance and reliability of the genetic retrieval enhancement model, reasonable architecture design, resource allocation and monitoring management are also required.

[0111] For 10 photovoltaic modules in a photovoltaic field, some faults were artificially created, such as blocking some photovoltaic modules, and the genetic retrieval enhancement model (Group A) and the genetic algorithm (Group B) were used for comparative experiments. The different processing methods of Group A and Group B were used to test the fault maintenance decision of the 10 modules, and the maintenance time and maintenance accuracy of the optimal maintenance decision schemes of Group A and Group B were compared. The comparison results are recorded in Table 3.

[0112] Table 3 Results Statistics

[0113]

[0114] According to experimental data, when testing the decision-making of a fault in a photovoltaic field, the maintenance decision-making scheme using the genetic retrieval enhanced model (Group A) is better than the scheme using the genetic algorithm (Group B). The maintenance time of Group A is 6.4 minutes less than that of Group B, and the maintenance accuracy is 20 percentage points higher, with higher efficiency and accuracy. This shows that the genetic retrieval enhanced model has better performance and application prospects in photovoltaic module fault maintenance.

[0115] The visual integrated management platform obtains the operating parameters of photovoltaic field equipment in real time through the high-speed serial expansion bus PCIe, including voltage, current, power, temperature, etc. Through interactive charts, heat maps, maps and dashboards, the changing trends, changing rules and equipment relationships of equipment data are monitored in real time. This allows managers and maintenance personnel to understand the operation of the equipment more intuitively and discover abnormal problems in time. The visual integrated management platform uses the Token user authentication mechanism to verify the identity of the accessing user. When accessing the platform, the user needs to provide an identity token. The platform will verify the validity of the user's identity based on the identity token and authorize the user to access the corresponding equipment data. After successful authentication, the user can remotely adjust the operating parameters of the photovoltaic field equipment by entering the event function, such as adjusting the value of certain parameters, or turning on and off certain equipment. This allows managers or maintenance personnel to manage equipment more safely and reliably when operating equipment remotely.

[0116] In general, the visual integrated management platform enables managers and maintenance personnel to more easily understand the operation of the equipment and discover abnormal problems in a timely manner by visually monitoring the equipment data. At the same time, the identity authentication mechanism adopted by the platform can effectively ensure the security of the equipment and prevent unauthorized users from operating the data. In addition, the platform also provides the function of remotely operating the equipment by inputting event functions, providing a more convenient operation method for managers or maintenance personnel.

[0117] The full-station acceleration communication network listens to the collected data forwarding method defined by the routing type through the three-layer TCP / UDP transmission protocol and the seven-layer HTTP / HTTPS cache protocol. By listening to the routing type definition, different types of data traffic can be accurately distinguished and processed accordingly. The full-station acceleration communication network distributes data traffic to the back-end server in real time, and realizes end-to-end transmission through the load balancing algorithm. The real-time distribution of data traffic can ensure the stability and fluency of data during transmission. The full-station acceleration communication network uses the low-voltage modulation communication standard LVM-hscs to perform differential Manchester encoding on the collected data to achieve data anti-interference communication. During the data transmission process, it may be interfered by the outside world, so the use of differential Manchester encoding can improve the stability of data transmission. The full-station acceleration communication network uses the secure socket layer protocol SSL acceleration card to encrypt and authenticate data transmission. During the data transmission process, data security protection is required to prevent data from being stolen or tampered with. Using SSL acceleration cards for data encryption and authentication can improve the security and confidentiality of data transmission.

[0118] In summary, the full-station acceleration communication network can ensure the stability and fluency of data during transmission through monitoring of routing type definition and real-time distribution of data traffic. At the same time, the use of differential Manchester encoding and SSL acceleration card for anti-interference communication and security protection of data can improve the stability, security and confidentiality of data transmission.

[0119] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. An intelligent operation and maintenance control system applied to a photovoltaic field, characterized in that: The intelligent operation and maintenance control system comprises: A real-time acquisition module, which collects the operating status and environmental parameters of photovoltaic field equipment in real time through a sensor network, and performs preliminary processing and analysis on the collected data by deploying edge computing nodes. The photovoltaic field equipment includes at least solar panels, inverters and battery energy storage equipment. The operating status and environmental parameters include at least current, voltage, temperature, power and light intensity. Photovoltaic cloud platform, which receives and stores edge processing data in real time through a full-station accelerated communication network, and performs distributed fault monitoring, diagnosis and early warning on the photovoltaic field equipment by building a lightweight twin fault supervision model; A fault maintenance module, which makes fault maintenance decisions through a genetic retrieval enhancement model, which adaptively generates a maintenance plan based on fault monitoring, diagnosis and early warning information, and performs fault maintenance control through remote control; An adjustment control module, wherein the adjustment control module intelligently controls and adjusts the operation of the photovoltaic field equipment through a fuzzy PID controller, and the fuzzy PID controller performs switch control, load control, power regulation control and battery energy storage management of the photovoltaic field equipment according to real-time operation data of the photovoltaic field, historical fault maintenance data and external environmental parameters; A multi-device management module, which realizes visual monitoring of the photovoltaic field equipment through a visual integrated management platform, and realizes centralized control and management of multiple photovoltaic field equipment nodes through a cloud server; The output end of the real-time acquisition module is connected to the input end of the photovoltaic cloud platform, the output end of the photovoltaic cloud platform is connected to the input end of the fault maintenance module, the output ends of the real-time acquisition module and the fault maintenance module are connected to the input end of the adjustment control module, the output end of the real-time acquisition module is connected to the input end of the multi-device management module, and the output end of the fault maintenance module is connected to the input end of the multi-device management module.

2. According to claim 1, an intelligent operation and maintenance control system applied to a photovoltaic field is characterized in that: The photovoltaic cloud platform builds a lightweight fault supervision model based on the physical information and dynamic sensor information of the photovoltaic field. The lightweight fault supervision model includes a physical entity layer, a virtual entity layer, a data interaction layer, a twin model layer, a model lightweight layer and an application layer. The physical entity layer drives the virtual entity to run in real time through real-time collected data. The virtual entity layer simulates the photovoltaic field subsystem, equipment component structure and dynamic operation status under different working conditions based on historical collected data and real-time collected data. The data interaction layer is used to transfer the data of the physical entity layer and the virtual entity layer to the twin model layer for processing. The twin model layer converts the actual data into real-time data through machine learning. The twin model layer performs feature matching and data comparison with the simulation data to realize fault monitoring, diagnosis and prediction of the equipment operation status. The model lightweight layer performs lightweight optimization on the lightweight fault supervision model through model pruning and parameter optimization. The application layer is used to summarize, integrate and interpret the output of the lightweight fault supervision model, and push alarm information and early warning information to management personnel or maintenance personnel. The twin model layer includes a fault detection module, a fault diagnosis module and a fault prediction module. The output end of the fault detection module is connected to the input end of the fault diagnosis module, and the output end of the fault diagnosis module is connected to the input end of the fault prediction module.

3. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 2, characterized in that: The fault detection module dynamically detects the power generation efficiency of the photovoltaic area based on the actual collected data and the simulated collected data, and detects the fault of the photovoltaic area by comparing the actual power generation efficiency of the photovoltaic area with the simulated power generation efficiency. The calculation formula of the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic area is: （1） In formula (1), represents the actual power generation efficiency of the photovoltaic field at the i-th sampling moment, r represents the actual photovoltaic field, represents the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, m represents the simulated photovoltaic field, Indicates the ith sampling instantaneous value of the actual output power of the photovoltaic module, , Represents the i-th sampling instantaneous value of the simulated output power of the photovoltaic module, represents the sampling period, Indicates the nominal power of the photovoltaic module. Indicates the instantaneous value of solar irradiance sampling on the front of the photovoltaic module. Indicates the solar irradiance under standard test conditions, , Represents the linear sampling parameter, which is used to adjust the deviation between the simulated output power and the actual output power of the photovoltaic module; the difference formula between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field is: （2） In formula (2), Indicates the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the i-th sampling moment, Indicates the maximum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum actual power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the maximum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, Indicates the minimum value of simulated power generation efficiency of the photovoltaic field at the moment of sampling, It is used to eliminate weight deviations caused by different data magnitudes and numerical ranges. When the difference between the actual power generation efficiency and the simulated power generation efficiency of the photovoltaic field at the sampling moment is greater than the similarity threshold, it is judged that the actual power generation efficiency of the photovoltaic field at the sampling moment is abnormal.

4. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 3 is characterized in that: The fault diagnosis module classifies and calculates the illumination, temperature, output voltage and current of the photovoltaic device and the working state of the inverter based on the energy efficiency difference between the photovoltaic components. First, the actual output power of the photovoltaic array is obtained according to the output voltage and current of the photovoltaic array. At the same time, the rated output power of the photovoltaic array in the current environment is calculated according to the light intensity, ambient temperature and photovoltaic array temperature, and the actual output power of the photovoltaic array is compared with the rated output power to obtain the conversion efficiency of the photovoltaic array. Then, the conversion efficiency of the inverter is calculated according to the input power and output power of the inverter, and the conversion efficiency of the inverter is monitored and diagnosed to determine whether the capacitor and the switch tube are faulty.

5. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 4, characterized in that: The fault prediction module collects the power timing curve data of the photovoltaic field equipment to extract the power generation efficiency and power output of the equipment in different time periods, and performs feature analysis on the power timing curve through time series analysis to discover potential inefficient power generation sites and issue early warnings.

6. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 1, characterized in that: The genetic retrieval enhancement model adaptively generates a maintenance plan based on a large model retrieval enhancement tool and a genetic algorithm. The working method of the genetic retrieval enhancement model includes the following steps: S1. Data input and initialization. The large model retrieval enhancement tool constructs a fault maintenance knowledge graph by retrieving the fault maintenance history records of the photovoltaic field and the expert knowledge base, and classifies the fault maintenance knowledge graph according to the equipment fault type and maintenance method, and preliminarily matches the fault monitoring, diagnosis and warning information with the fault maintenance knowledge graph to generate an initial fault maintenance strategy population; S2, population evolution, evaluates and ranks the solutions in the initial population through the fitness function, and evolves the initial population through population individual selection, crossover and mutation operations; S3, dynamically updating parameters, the large model retrieval enhancement tool dynamically updates the fitness function, selection strategy, crossover and mutation operations of the genetic algorithm according to the real-time collected data and the retrieval maintenance decision; S4, iteration and convergence judgment, judging whether the maximum number of iterations or fitness value convergence is reached, if the maximum number of iterations or fitness value convergence is reached, then execute S5 operation, otherwise execute S2 operation; S5. Output the optimal maintenance decision plan, and feed back the data and decision results obtained during the execution process to the fault maintenance knowledge graph for updating.

7. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 1, characterized in that: The visual integrated management platform obtains the operating parameters of the photovoltaic field equipment in real time through the high-speed serial expansion bus PCIe, and uses interactive charts, heat maps, maps and dashboards to perform real-time visual monitoring of equipment data change trends, change rules and equipment relationships. The visual integrated management platform uses a Token user identity authentication mechanism to verify the identity of the accessing user. After the identity authentication is successful, the user can remotely adjust the operating parameters of the photovoltaic field equipment by inputting event functions.

8. The intelligent operation and maintenance control system applied to a photovoltaic field according to claim 1, characterized in that: The whole-station acceleration communication network listens to the collected data forwarding method defined by the routing type through the three-layer TCP / UDP transmission protocol and the seven-layer HTTP / HTTPS cache protocol, and distributes the data traffic to the back-end server in real time to achieve end-to-end transmission. The whole-station acceleration communication network uses the low-voltage modulation communication standard LVM-hscs to perform differential Manchester encoding on the collected data to achieve data anti-interference communication, and uses the Secure Sockets Layer protocol SSL acceleration card to encrypt and authenticate data transmission.

Images