A photovoltaic power station cleaning process evaluation method and system based on matrix analysis

Abstract

This invention proposes a method and system for evaluating the cleaning process of photovoltaic (PV) power plants based on matrix analysis. The method includes: real-time acquisition of operating parameter information of the PV power plant and transmission of this information to a central system; classifying PV modules into different region matrices using predetermined classification criteria; and training a deep learning model using the region matrix information combined with historical performance data of the PV power plant to obtain a trained and optimized deep learning model. By real-time monitoring and accurate identification of the cleaning needs of PV modules, and by optimizing the cleaning schedule, this invention can maximize the power generation efficiency of PV power plants. This helps to fully utilize solar energy resources and provide more clean energy.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station cleaning process evaluation, and in particular to a photovoltaic power station cleaning process evaluation method and system based on matrix analysis. Background Art

[0002] Photovoltaic power station cleaning process assessment technology is one of the most concerned and rapidly developing fields in the solar photovoltaic industry. Its importance lies in ensuring the efficient operation of solar photovoltaic modules, maximizing power generation efficiency through real-time monitoring and accurate identification of cleaning needs, while reducing maintenance costs, thereby promoting the sustainable development of renewable energy.

[0003] At present, there are some problems and shortcomings in the traditional methods for evaluating the cleaning process of photovoltaic power plants:

[0004] 1. Fixed cleaning schedule: Traditional methods usually perform cleaning based on a fixed schedule, rather than developing a cleaning schedule based on actual needs and the condition of the PV panels. This may result in insufficient or excessive cleaning, wasting resources and reducing power generation efficiency.

[0005] 2. Cleaning force allocation problem: Traditional methods often fail to optimize cleaning force and fail to identify which components or areas need more urgent cleaning. This may cause some severely polluted areas to be ignored, thus affecting the performance of the entire power station.

[0006] 3. Lack of real-time monitoring: Traditional methods usually cannot provide real-time monitoring and feedback, so problems such as poor cleaning effect or local dust accumulation cannot be discovered in time. This delays the identification and resolution of problems and reduces power generation efficiency.

[0007] 4. Accurate dust accumulation model: The traditional dust accumulation prediction model is calculated based on the overall historical data of the power station, lacking classification and prior knowledge, resulting in poor accuracy of the dust accumulation model.

[0008] The matrix analysis model-based approach can consider multiple factors and situations more comprehensively, and therefore has the potential to become a key tool for the evaluation technology of the cleaning process of photovoltaic power plants. Summary of the invention

[0009] The present invention provides a photovoltaic power station cleaning process evaluation method and system based on matrix analysis, which is used to solve the problems in the prior art that the cleaning method has low cleaning efficiency, cannot monitor the cleaning effect in real time, mostly relies on manual inspection, has high human resource cost, cannot dynamically adjust the cleaning strategy, resulting in poor cleaning effect, sampling cleaning robots lead to high cost, and cannot accurately evaluate the cleanliness of photovoltaic components, and cannot carry out regional cleaning plans or in-depth cleaning effect evaluation:

[0010] The present invention proposes a photovoltaic power station cleaning process evaluation method based on matrix analysis, and the photovoltaic power station cleaning process evaluation method comprises:

[0011] S1: Collecting operating parameter information of the photovoltaic power station in real time and sending the operating parameter information to the central system;

[0012] S2: Divide the photovoltaic modules into different regional matrices using a predetermined classification standard;

[0013] S3: training a deep learning model using the regional matrix information of the regional matrix combined with historical performance data of the photovoltaic power station to obtain a trained and optimized deep learning model;

[0014] S4: Determine the strings of the photovoltaic power station that need to be cleaned and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements;

[0015] S5: Evaluate the cleaning effect based on the evaluation results of the strings.

[0016] Furthermore, real-time collection of operating parameter information of the photovoltaic power station and sending the operating parameter information to the central system include:

[0017] S11: Controlling the data acquisition device to regularly measure other performance parameters such as current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station;

[0018] S12: Send the operating parameter information to the central system through the communication module.

[0019] Furthermore, the photovoltaic modules are divided into different regional matrices using predetermined classification standards, including:

[0020] S21: extracting the predetermined classification standard;

[0021] S22: using a matrix analysis method to divide the photovoltaic modules according to the predetermined classification standard to form different regional matrices;

[0022] S23: Classifying the photovoltaic modules into corresponding regional matrices according to the division results.

[0023] Furthermore, according to the cleaning requirements, the strings of the photovoltaic power station that need to be cleaned and their cleaning parameters are determined by the trained and optimized deep learning model, including:

[0024] S41: Retrieving the cleaning requirements of the photovoltaic power station;

[0025] S42: Inputting the cleaning requirement into the trained and optimized deep learning model;

[0026] S43: inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model;

[0027] S44: for the strings that need to be cleaned, the trained and optimized deep learning model is used to determine cleaning parameters according to the operation parameters of the strings, wherein the cleaning parameters include cleaning sequence, cleaning intensity and cleaning duration;

[0028] S45: performing cleaning intensity allocation and cleaning duration allocation according to the cleaning intensity and cleaning duration of each string to be cleaned, and performing cleaning scheduling for the strings to be cleaned according to the cleaning sequence.

[0029] Furthermore, the cleaning effect is evaluated based on the evaluation results of the string, including:

[0030] S51: real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree;

[0031] S52: Evaluate the cleaning degree in combination with the regional matrix information;

[0032] S53: Evaluate the power generation efficiency of the photovoltaic module using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data to obtain an evaluation result, and determine whether the photovoltaic module is in the best performance state based on the evaluation result.

[0033] The present invention proposes a photovoltaic power station cleaning process evaluation system based on matrix analysis, and the photovoltaic power station cleaning process evaluation system includes:

[0034] Information collection module: collects the operating parameter information of the photovoltaic power station in real time and sends the operating parameter information to the central system;

[0035] Matrix division module: divides photovoltaic modules into different regional matrices using predetermined classification standards;

[0036] Model acquisition module: using the regional matrix information of the regional matrix combined with the historical performance data of the photovoltaic power station to train the deep learning model, and obtain a trained and optimized deep learning model;

[0037] Parameter determination module: determining the strings that need to be cleaned in the photovoltaic power station and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements;

[0038] Effect evaluation module: Evaluate the cleaning effect based on the evaluation results of the group strings.

[0039] Furthermore, the information collection module includes:

[0040] Parameter measurement module: controls the data acquisition device to regularly measure other performance parameters such as current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station;

[0041] Information transmission module: sends the operating parameter information to the central system through the communication module.

[0042] Furthermore, the matrix partitioning module includes:

[0043] Standard extraction module: extracting the predetermined classification standard;

[0044] Component division module: using a matrix analysis method to divide photovoltaic components according to the predetermined classification standard to form different regional matrices;

[0045] Component classification module: classifies the photovoltaic components into their corresponding regional matrices according to the division results.

[0046] Furthermore, the parameter determination module includes:

[0047] Retrieving the cleaning requirements of the photovoltaic power station;

[0048] Demand input module: inputs the cleaning demand into the trained and optimized deep learning model;

[0049] Demand determination module: inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model;

[0050] Parameter determination module: for the strings that need to be cleaned, the cleaning parameters are determined according to the operating parameters of the strings using the trained and optimized deep learning model, wherein the cleaning parameters include cleaning sequence, cleaning intensity and cleaning duration;

[0051] Cleaning scheduling module: according to the cleaning intensity and cleaning time of each string to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning of the strings to be cleaned is scheduled according to the cleaning order.

[0052] Furthermore, the effect evaluation module includes:

[0053] Real-time monitoring module: real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree;

[0054] Degree evaluation module: evaluates the cleaning degree in combination with the regional matrix information;

[0055] State judgment module: evaluate the power generation efficiency of the photovoltaic module using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data, obtain the evaluation results, and judge whether the photovoltaic module is in the best performance state through the evaluation results.

[0056] Beneficial effects of the present invention:

[0057] 1. Improve power generation efficiency: By real-time monitoring and accurate identification of the cleaning needs of photovoltaic modules, as well as optimizing the cleaning schedule, the present invention can maximize the power generation efficiency of photovoltaic power stations. This helps to fully utilize solar energy resources and provide more clean energy.

[0058] 2. Reduce maintenance costs: The present invention uses deep learning models to accurately identify which photovoltaic components need to be cleaned and which do not, as well as the optimal cleaning sequence. This effectively reduces unnecessary cleaning work, reduces the use of cleaning equipment and manpower, and thus reduces maintenance costs.

[0059] 3. Real-time cleaning effect monitoring: The present invention introduces cleaning degree sensors and matrix analysis to monitor the cleaning status of photovoltaic modules in real time and provide cleaning effect evaluation through deep learning models. This enables operation and maintenance personnel to promptly discover incompletely cleaned modules or local dust accumulation problems, ensuring the real-time effect of the cleaning process.

[0060] 4. Promote the sustainable development of renewable energy: By improving the power generation efficiency of photovoltaic power stations, reducing maintenance costs, and reducing dependence on fossil fuels, the present invention helps promote the sustainable development of renewable energy (solar energy) and reduce the impact on the environment.

[0061] 5. Intelligent maintenance decision-making: The present invention establishes an intelligent cleaning intensity allocation model based on deep learning and historical data, enabling the system to automatically identify cleaning needs and allocate cleaning resources, providing operation and maintenance personnel with more intelligent maintenance decision support.

[0062] 6. Reduce human errors: Traditional manual maintenance methods may be affected by human errors, while the data and model-based method of the present invention reduces such errors and improves the accuracy and efficiency of cleaning. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1A step diagram of a photovoltaic power station cleaning process evaluation method based on matrix analysis proposed by the present invention;

[0064] Figure 2 This is a flow chart of a photovoltaic power station cleaning process evaluation system based on matrix analysis proposed by the present invention. DETAILED DESCRIPTION

[0065] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0066] Embodiment 1,

[0067] This embodiment proposes a photovoltaic power station cleaning process evaluation method based on matrix analysis, such as Figure 1 As shown, the photovoltaic power station cleaning process evaluation method includes:

[0068] S1: Collecting operating parameter information of the photovoltaic power station in real time and sending the operating parameter information to the central system;

[0069] S2: Divide the photovoltaic modules into different regional matrices using a predetermined classification standard;

[0070] S3: training a deep learning model using the regional matrix information of the regional matrix combined with historical performance data of the photovoltaic power station to obtain a trained and optimized deep learning model;

[0071] S4: Determine the photovoltaic power station strings that need to be cleaned and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements;

[0072] S5: Evaluate the cleaning effect based on the evaluation results of the strings.

[0073] The working principle of the above technical solution is: real-time collection of operating parameter information of the photovoltaic power station, and sending the operating parameter information to the central system. By monitoring the operating parameters of the photovoltaic power station, such as power generation power, voltage, temperature, etc., the real-time status and performance data of the power station can be obtained; the photovoltaic components are divided into different regional matrices using predetermined classification standards. According to the location, orientation, inclination and other characteristics of the photovoltaic components, the photovoltaic components are divided into different regions, and each region forms a matrix; the regional matrix information of the regional matrix is ​​combined with the historical performance data of the photovoltaic power station to train the deep learning model to obtain a deep learning model that has been trained and optimized. The regional matrix information and historical performance data are used as training samples, and the model is trained and optimized by a deep learning algorithm to establish a model for evaluating the cleaning process of the photovoltaic power station; according to the cleaning requirements, the strings that need to be cleaned in the photovoltaic power station and their cleaning parameters are determined by the deep learning model that has been trained and optimized. According to the cleaning requirements of the photovoltaic power station, the deep learning model that has been trained and optimized is used to evaluate each area, determine the strings that need to be cleaned and the corresponding cleaning parameters, such as cleaning time, cleaning agent concentration, etc.; and the cleaning effect is evaluated according to the evaluation results of the strings. After cleaning is completed, the cleaning effect is evaluated and analyzed by monitoring the operating parameters and performance data of the PV modules to determine whether the cleaning process has achieved the expected results. The cleaning strategy and parameters can be adjusted based on the evaluation results.

[0074] Specifically, the real-time monitoring system corresponding to the real-time collection of operating parameter information of the photovoltaic power station may include:

[0075] Data acquisition device: The data acquisition device can collect key information such as voltage, current, string current, light intensity, etc. of the photovoltaic power station.

[0076] Data acquisition unit: responsible for collecting data generated by the data acquisition device. This unit regularly collects the operating data of the photovoltaic power station.

[0077] Communication module: connected to the data acquisition unit, used to transmit the collected data to the central system. The communication module can use wireless or wired connection to ensure real-time data transmission.

[0078] The data acquisition device regularly measures performance parameters such as current, voltage, temperature, etc. The data acquisition unit is responsible for regularly collecting the photovoltaic power station operation data collected by the data acquisition device and transmitting it to the central system.

[0079] The actual structure of the central system may include:

[0080] Computer or Server: The core component of the central system, used for data processing and analysis. It is connected to the communication module and receives data from the real-time monitoring system.

[0081] Matrix analysis model: The central system includes a matrix analysis model, which is used to divide photovoltaic modules into different regional matrices for regional classification. The model classifies modules based on mechanism and data analysis criteria.

[0082] Deep Learning Model: Used to build and run a deep learning model that analyzes historical performance data, predicts the theoretical optimal power generation for each region, and learns the historical optimal power generation efficiency.

[0083] Cleaning intensity allocation algorithm: The central system is equipped with a cleaning intensity allocation algorithm, which determines which strings need to be cleaned and which do not, based on the output of the deep learning model. The algorithm ensures that the strings that need cleaning the most receive sufficient cleaning intensity.

[0084] Scheduling Optimization: The central system also includes a scheduling optimization algorithm, which determines the working order and time allocation of the cleaning robots based on the distribution of cleaning intensity to maximize power generation efficiency.

[0085] After receiving the real-time monitoring data, the central system starts the matrix analysis model. The matrix analysis model determines the classification criteria based on mechanism analysis and data cluster analysis, and divides the photovoltaic modules into different regional matrices, such as orientation, bracket type, etc.

[0086] Components are grouped into their own regional matrix, forming a regionalized classification.

[0087] Furthermore, the connection between the real-time monitoring system and the central system is as follows:

[0088] The data acquisition device is connected to the data acquisition unit via a cable to achieve real-time data transmission.

[0089] The data acquisition unit is connected to the communication module via an internal data bus or a communication protocol so as to transmit the collected data to the central system.

[0090] The computer or server in the central system is connected to the communication module to receive, store and process data from the real-time monitoring system.

[0091] The matrix analysis model, deep learning model, cleaning intensity allocation algorithm and scheduling optimization algorithm within the central system work together to analyze and make decisions based on the data received.

[0092] Among them, the deep learning model uses a deep convolutional neural network for modeling. The input data is the time series of output power / installed capacity, string current, solar radiation intensity, temperature, and wind speed; the output data is the cleanliness of the components.

[0093] In order to capture the temporal information of the data, the data is processed by dynamic windows to form a dynamic matrix, which serves as the input matrix of the deep neural network.

[0094] The network structure of the model is:

[0095] The first layer is convolution, and the convolution kernel is 3×3

[0096] The second layer is pooling, using 2×2 maximum pooling

[0097] The third layer is convolution, and the convolution kernel is 3×3

[0098] The fourth layer is pooling, using 2×2 maximum pooling

[0099] Finally, there are 3 fully connected layers.

[0100] The activation function of the entire model uses ReLU.

[0101] The effects of the above technical solution are as follows: by collecting the operating parameter information of the photovoltaic power station in real time, the real-time status and performance data of the power station can be obtained in time, and the cleaning needs of the photovoltaic modules can be accurately evaluated; by using the predetermined classification standards to divide the photovoltaic modules into regional matrices, differentiated cleaning strategies can be formulated according to different regional characteristics to improve the cleaning effect; by using the regional matrix information and historical performance data to train and optimize the deep learning model, an accurate evaluation model can be established to improve the accuracy and reliability of the cleaning evaluation; through the trained and optimized deep learning model, the photovoltaic power station strings that need to be cleaned and the corresponding cleaning parameters, such as cleaning time, cleaning agent concentration, etc., can be determined according to the cleaning needs, thereby ensuring the efficiency and controllability of the cleaning process; according to the evaluation results of the strings, the cleaning effect can be evaluated to determine whether the cleaning process has achieved the expected effect, and the cleaning strategy and parameters can be adjusted in time to ensure the normal operation and power generation efficiency of the photovoltaic modules.

[0102] Embodiment 2,

[0103] In this embodiment, real-time collection of operating parameter information of a photovoltaic power station and sending the operating parameter information to a central system include:

[0104] S11: Controlling the data acquisition device to regularly measure other performance parameters such as current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station;

[0105] S12: Send the operating parameter information to the central system through the communication module.

[0106] The working principle of the above technical solution is: install a data acquisition device in the photovoltaic power station to regularly measure the current, voltage, temperature, light intensity and other performance parameters of each string. These parameters are important indicators that affect the power generation efficiency of the photovoltaic power station. Through real-time monitoring, the operating status of the power station can be timely grasped; the data acquisition device needs to perform regular measurements to ensure the accuracy and reliability of the data. The control data acquisition device can collect and upload data regularly to avoid errors and delays in manual operation; the operating parameter information is sent to the central system through the communication module to uniformly manage and analyze the data. The communication module can use wireless communication or wired communication to achieve remote monitoring and management of data; the central system is responsible for receiving and processing the operating parameter information, analyzing and processing the data, analyzing the operating status and performance of the photovoltaic power station, and proposing corresponding suggestions and measures. The central system can also be linked with other systems, such as alarm systems, maintenance systems, etc., to achieve comprehensive monitoring and management of photovoltaic power stations.

[0107] The effects of the above technical solution are as follows: by collecting the operating parameter information of the photovoltaic power station in real time, the real-time status and performance data of the power station can be obtained in time, the operating status of the photovoltaic components can be accurately evaluated, and faults and abnormal conditions can be discovered in time; by analyzing the operating parameter information, the operating status and performance of the photovoltaic power station can be understood, and data support can be provided for the optimized management of the photovoltaic components to improve the power generation efficiency and operating reliability of the photovoltaic power station; by monitoring and analyzing the operating parameter information, faults and abnormal conditions in the power station can be discovered in time, and accurate fault diagnosis and elimination can be carried out to avoid power station shutdown and losses caused by faults; by analyzing the historical data of the operating parameter information, the future operating status and performance of the photovoltaic power station can be predicted, reasonable maintenance planning and preventive measures can be carried out, and the operating time and benefits of the power station can be improved; by monitoring and analyzing the operating parameter information, the management and operation of the photovoltaic power station can be optimized, scientific management measures and strategies can be formulated, and the operating efficiency and economic benefits of the photovoltaic power station can be improved.

[0108] Embodiment 3,

[0109] In this embodiment, the photovoltaic modules are divided into different regional matrices using a predetermined classification standard, including:

[0110] S21: extracting the predetermined classification standard;

[0111] S22: using a matrix analysis method to divide the photovoltaic modules according to the predetermined classification standard to form different regional matrices;

[0112] S23: Classifying the photovoltaic modules into corresponding regional matrices according to the division results.

[0113] The working principle of the above technical solution is: extracting predetermined classification standards, which can be parameters such as size, material, color, thickness, etc. of photovoltaic modules, or indicators such as electrical performance, optical performance, mechanical performance, etc.; using matrix analysis method, taking the extracted classification standards as dimensions, performing data analysis and matrix operations on different dimensions, and obtaining different regional matrices. These regional matrices can be composed of multiple attributes to form a multidimensional attribute space, and different regional divisions can be obtained by combining different attributes; according to the classification results, photovoltaic modules are classified into corresponding regional matrices according to their characteristics and attributes to realize the classification management and refined management of photovoltaic modules.

[0114] The division of the regional matrix is ​​specifically as follows:

[0115] The division of the regional matrix adopts a method combining mechanism analysis and cluster analysis; when there is rich prior knowledge, the mechanism analysis is as follows:

[0116] 1) The regions are divided according to the roof of the photovoltaic power station, that is, the components on the same roof are divided into 1 region;

[0117] 2) The photovoltaic modules are divided into different areas according to their orientation and inclination, i.e. modules with the same orientation and inclination are in the same area.

[0118] When detailed information on the photovoltaic power plant site is not available:

[0119] Division based on data clustering method:

[0120] The k-nearest neighbor method is used to cluster the power / installed capacity, solar radiation intensity, wind speed, and temperature, and the photovoltaic power station areas are divided to ensure that the data characteristics of components in the same area are more similar.

[0121] The effect of the above technical solution is: by dividing the photovoltaic modules according to the predetermined classification standards, different types of photovoltaic modules can be managed in a refined manner. In this way, the performance characteristics and operating status of different types of photovoltaic modules can be better understood and mastered, and management strategies and maintenance plans can be formulated in a targeted manner; by dividing the photovoltaic modules into different regional matrices, more accurate data analysis and management can be achieved. According to the characteristics and properties of the photovoltaic modules in different regional matrices, targeted optimization and adjustment can be carried out to improve the power generation efficiency and performance of the photovoltaic modules; refined management and optimization adjustment can help identify and deal with problems and faults in photovoltaic modules. By timely discovering and solving problems, the cost of maintenance and replacement can be reduced, the service life of photovoltaic modules can be extended, and thus the operating cost can be reduced; by dividing the photovoltaic modules into regional matrices, the quality status of photovoltaic modules can be better grasped. For photovoltaic modules in different regional matrices, targeted quality control and monitoring measures can be taken to improve the overall quality level of photovoltaic modules; by analyzing the data of photovoltaic modules in the regional matrix, more accurate and comprehensive information can be obtained. This information can provide reference and support for decision makers, help them make scientific and reasonable operation and management decisions, and improve the overall benefits of photovoltaic power stations.

[0122] Embodiment 4,

[0123] In this embodiment, the strings of the photovoltaic power station that need to be cleaned and their cleaning parameters are determined according to the cleaning requirements through the trained and optimized deep learning model, including:

[0124] S41: Retrieving the cleaning requirements of the photovoltaic power station;

[0125] S42: Inputting the cleaning requirement into the trained and optimized deep learning model;

[0126] S43: inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model;

[0127] S44: For the strings that need to be cleaned, the trained and optimized deep learning model is used to determine cleaning parameters according to the operation parameters of the strings, wherein the cleaning parameters include cleaning sequence, cleaning intensity and cleaning duration; specifically, the cleaning intensity allocation process is as follows:

[0128] Using historical operating data, input data including string current, string voltage, inverter power, light intensity, temperature, time, etc., a deep learning model is established in the central system.

[0129] The model predicts the theoretical optimal power generation for each region and learns the historical optimal power generation efficiency.

[0130] Adam’s method is used to optimize the model parameters to improve accuracy and performance.

[0131] Using deep learning models to learn cleaning requirements, the system analyzes the performance parameters and cleaning degree data of each string, as well as the calculated component dust and dust accumulation levels.

[0132] The system determines which strings need to be cleaned and which do not, ensuring that the strings that need cleaning the most receive the cleaning strength. The specific strength ratio of the cleaning strength is obtained through the following formula:

[0133]

[0134] in, P A specific intensity ratio indicating the intensity of cleaning; P 0 represents the preset ratio reference value (i.e., the initialization ratio); M represents the dust parameter matrix; M 0 represents the benchmark parameter matrix; m h and m hs Respectively represent the dust mass and dust distribution area of ​​the component during the current cleaning; Δ m h and Δ m hs They respectively represent the mass change of the component dust and the dust distribution area change between the current cleaning and the last cleaning; m hp and m hps The preset component dust mass reference value and dust distribution area reference value corresponding to the ratio reference value are represented respectively;

[0135] The cleaning force is obtained by using the specific force ratio of the cleaning force, wherein the cleaning force is obtained by the following formula:

[0136]

[0137] in, F Indicates the intensity of cleaning; F 0 represents the preset initial value of cleaning intensity;

[0138] At the same time, the cleaning time is obtained by the following formula:

[0139]

[0140] in, T Indicates the cleaning time; T 0 indicates the preset initial value of cleaning time;

[0141] S45: performing cleaning intensity allocation and cleaning duration allocation according to the cleaning intensity and cleaning duration of each string to be cleaned, and performing cleaning scheduling for the strings to be cleaned according to the cleaning sequence.

[0142] The working principle of the above technical solution is: by retrieving the cleaning requirements of the photovoltaic power station, it is input into the deep learning model that has been trained and optimized. Then, the operating parameters of each string of the photovoltaic power station are input into the deep learning model, and the deep learning model is used to analyze and judge the input operating parameters to determine whether each string needs to be cleaned. For the strings that need to be cleaned, the deep learning model is used to determine the cleaning parameters according to their operating parameters, including the cleaning order, cleaning intensity and cleaning time. Finally, according to the cleaning intensity and cleaning time of each string that needs to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning schedule of the strings to be cleaned is carried out according to the determined cleaning order.

[0143] The effect of the above technical solution is: by completing the training and optimization of the deep learning model, it is possible to accurately determine whether each group of strings in the photovoltaic power station meets the cleaning requirements. Based on the analysis and judgment of the deep learning model, the error of subjective judgment can be eliminated and the accuracy of the cleaning judgment can be improved; using the deep learning model, the entire cleaning process can be automatically executed, reducing the need for manual operation. Only by inputting the cleaning requirements of the photovoltaic power station and the operating parameters of each group of strings, the deep learning model can automatically judge and determine the cleaning parameters, greatly improving the efficiency and convenience of the cleaning process; by completing the training and optimization of the deep learning model, the cleaning parameters can be determined according to the operating parameters of the group of strings, including the cleaning sequence, cleaning intensity and cleaning duration. In this way, personalized cleaning treatment can be carried out for different groups of strings, maximizing the cleaning effect and ensuring the normal operation and power generation efficiency of the photovoltaic power station; due to the precise determination of the cleaning parameters, the cleaning intensity and duration can be allocated according to actual needs, avoiding unnecessary cleaning waste and resource consumption, thereby saving cleaning costs. At the same time, automated cleaning scheduling can optimize the use of cleaning time and human resources, further reducing the cost of cleaning.

[0144] On the other hand, by obtaining the dust parameter matrix and the preset component dust mass baseline value and dust distribution area baseline value, the degree of contamination and distribution of the component can be judged more accurately, so as to adjust the cleaning intensity and cleaning time according to the actual situation, and effectively improve the cleaning quality. By monitoring the change in the mass of the component dust and the change in the dust distribution area between the current cleaning and the last cleaning, the wear of the component can be discovered in time, so as to adjust the cleaning intensity and cleaning time, reduce damage to the component, and extend the service life of the component. Through the preset initial value of the cleaning intensity and the preset initial value of the cleaning time, reasonable settings can be made according to actual needs to avoid excessive cleaning and energy waste, and realize the effective use of energy. Through precise control of the cleaning intensity, the cleaning process can be automated and intelligently managed, reducing manual intervention and operational errors, and improving work efficiency.

[0145] In summary, this control method based on cleaning intensity can achieve good technical results in improving cleaning quality, extending component life, saving energy and improving work efficiency.

[0146] Embodiment 5,

[0147] In this embodiment, the cleaning effect is evaluated according to the evaluation result of the string, including:

[0148] S51: real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree;

[0149] S52: Evaluate the cleaning degree in combination with the regional matrix information;

[0150] S53: Evaluate the power generation efficiency of the photovoltaic module using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data to obtain an evaluation result, and determine whether the photovoltaic module is in the best performance state based on the evaluation result.

[0151] The working principle of the above technical solution is: real-time monitoring of data parameters during string cleaning, including performance parameters and cleaning degree. These parameters can be monitored and collected in real time by sensors and other equipment to reflect the actual situation of the string. The cleaning degree is evaluated in combination with regional matrix information. Regional matrix is ​​a commonly used evaluation method in photovoltaic power stations. The effect of string cleaning can be obtained by comparing the differences in power generation efficiency of components in different regions. Using regional matrix information, the cleaning degree of the string can be objectively evaluated. The power generation efficiency of photovoltaic components is evaluated by using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data. The deep learning model can predict the power generation efficiency of the component through historical data and real-time monitoring data, and compare it with the historical optimal power generation efficiency to obtain the current power generation efficiency of the component. According to the evaluation results, it is judged whether the photovoltaic component is in the best performance state. According to the real-time monitoring data and evaluation results, it can be judged whether the component has reached the best performance state. If not, further cleaning or other treatment is required.

[0152] At the same time, the model is established using the deep convolutional neural network method, which will output the cleanliness of the components of each matrix. A model will be established for each matrix, and RMSE and MAE will be used to evaluate the model accuracy. If the model performance meets the parameter requirements, the cleanliness obtained by the model can be used for reference.

[0153] Through the deep learning model, the changes in the cleanliness level of each matrix area can be seen in real time. When the cleanliness level reaches the preset threshold, it means that the matrix area is in a clean state. The entire process can show the dynamic changes in the cleanliness level of each matrix in the photovoltaic power station, thereby realizing the evaluation of the cleanliness level of the matrix.

[0154] The effects of the above technical solution are: by real-time monitoring of the performance parameters and cleaning degree during the string cleaning process, the operation status of the string can be obtained in time to ensure accurate evaluation and maintenance of the photovoltaic modules; combined with the regional matrix information, the cleaning degree can be evaluated. In this way, the cleaning effect of the string can be judged more objectively to avoid performance degradation or damage caused by incomplete or excessive cleaning; the power generation efficiency of the photovoltaic module is evaluated using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data. By comparing with the historical optimal power generation efficiency, the current power generation efficiency of the module can be understood and abnormal conditions can be discovered in time; through the evaluation results, it can be determined whether the photovoltaic module is in the best performance state. If the evaluation results show that the module is not in the best state, corresponding measures can be taken, such as further cleaning, repair or replacement of the module, to ensure the normal operation and power generation efficiency of the photovoltaic power station.

[0155] Embodiment 6,

[0156] This embodiment proposes a photovoltaic power station cleaning process evaluation system based on matrix analysis, such as Figure 2 As shown, the photovoltaic power station cleaning process evaluation system includes:

[0157] Information collection module: collects the operating parameter information of the photovoltaic power station in real time and sends the operating parameter information to the central system;

[0158] Matrix division module: divides photovoltaic modules into different regional matrices using predetermined classification standards;

[0159] Model acquisition module: using the regional matrix information of the regional matrix combined with the historical performance data of the photovoltaic power station to train the deep learning model, and obtain a trained and optimized deep learning model;

[0160] Parameter determination module: determining the strings that need to be cleaned in the photovoltaic power station and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements;

[0161] Effect evaluation module: Evaluate the cleaning effect based on the evaluation results of the group strings.

[0162] The working principle of the above technical solution is: real-time collection of operating parameter information of the photovoltaic power station, and sending the operating parameter information to the central system. By monitoring the operating parameters of the photovoltaic power station, such as power generation power, voltage, temperature, etc., the real-time status and performance data of the power station can be obtained; the photovoltaic components are divided into different regional matrices using predetermined classification standards. According to the location, orientation, inclination and other characteristics of the photovoltaic components, the photovoltaic components are divided into different regions, and each region forms a matrix; the regional matrix information of the regional matrix is ​​combined with the historical performance data of the photovoltaic power station to train the deep learning model to obtain a deep learning model that has been trained and optimized. The regional matrix information and historical performance data are used as training samples, and the model is trained and optimized by a deep learning algorithm to establish a model for evaluating the cleaning process of the photovoltaic power station; according to the cleaning requirements, the strings that need to be cleaned in the photovoltaic power station and their cleaning parameters are determined by the deep learning model that has been trained and optimized. According to the cleaning requirements of the photovoltaic power station, the deep learning model that has been trained and optimized is used to evaluate each area, determine the strings that need to be cleaned and the corresponding cleaning parameters, such as cleaning time, cleaning agent concentration, etc.; and the cleaning effect is evaluated according to the evaluation results of the strings. After cleaning is completed, the cleaning effect is evaluated and analyzed by monitoring the operating parameters and performance data of the PV modules to determine whether the cleaning process has achieved the expected results. The cleaning strategy and parameters can be adjusted based on the evaluation results.

[0163] The effects of the above technical solution are as follows: by collecting the operating parameter information of the photovoltaic power station in real time, the real-time status and performance data of the power station can be obtained in time, and the cleaning needs of the photovoltaic modules can be accurately evaluated; by using the predetermined classification standards to divide the photovoltaic modules into regional matrices, differentiated cleaning strategies can be formulated according to different regional characteristics to improve the cleaning effect; by using the regional matrix information and historical performance data to train and optimize the deep learning model, an accurate evaluation model can be established to improve the accuracy and reliability of the cleaning evaluation; through the trained and optimized deep learning model, the photovoltaic power station strings that need to be cleaned and the corresponding cleaning parameters, such as cleaning time, cleaning agent concentration, etc., can be determined according to the cleaning needs, thereby ensuring the efficiency and controllability of the cleaning process; according to the evaluation results of the strings, the cleaning effect can be evaluated to determine whether the cleaning process has achieved the expected effect, and the cleaning strategy and parameters can be adjusted in time to ensure the normal operation and power generation efficiency of the photovoltaic modules.

[0164] Embodiment 7,

[0165] In this embodiment, the information collection module includes:

[0166] Parameter measurement module: controls the data acquisition device to regularly measure other performance parameters such as current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station;

[0167] Information transmission module: sends the operating parameter information to the central system through the communication module.

[0168] The working principle of the above technical solution is: install a data acquisition device in the photovoltaic power station to regularly measure the current, voltage, temperature, light intensity and other performance parameters of each string. These parameters are important indicators that affect the power generation efficiency of the photovoltaic power station. Through real-time monitoring, the operating status of the power station can be timely grasped; the data acquisition device needs to perform regular measurements to ensure the accuracy and reliability of the data. The control data acquisition device can collect and upload data regularly to avoid errors and delays in manual operation; the operating parameter information is sent to the central system through the communication module to uniformly manage and analyze the data. The communication module can use wireless communication or wired communication to achieve remote monitoring and management of data; the central system is responsible for receiving and processing the operating parameter information, analyzing and processing the data, analyzing the operating status and performance of the photovoltaic power station, and proposing corresponding suggestions and measures. The central system can also be linked with other systems, such as alarm systems, maintenance systems, etc., to achieve comprehensive monitoring and management of photovoltaic power stations.

[0169] The effects of the above technical solution are as follows: by collecting the operating parameter information of the photovoltaic power station in real time, the real-time status and performance data of the power station can be obtained in time, the operating status of the photovoltaic components can be accurately evaluated, and faults and abnormal conditions can be discovered in time; by analyzing the operating parameter information, the operating status and performance of the photovoltaic power station can be understood, and data support can be provided for the optimized management of the photovoltaic components to improve the power generation efficiency and operating reliability of the photovoltaic power station; by monitoring and analyzing the operating parameter information, faults and abnormal conditions in the power station can be discovered in time, and accurate fault diagnosis and elimination can be carried out to avoid power station shutdown and losses caused by faults; by analyzing the historical data of the operating parameter information, the future operating status and performance of the photovoltaic power station can be predicted, reasonable maintenance planning and preventive measures can be carried out, and the operating time and benefits of the power station can be improved; by monitoring and analyzing the operating parameter information, the management and operation of the photovoltaic power station can be optimized, scientific management measures and strategies can be formulated, and the operating efficiency and economic benefits of the photovoltaic power station can be improved.

[0170] Embodiment 8,

[0171] In this embodiment, the matrix division module includes:

[0172] Standard extraction module: extracting the predetermined classification standard;

[0173] Component division module: using a matrix analysis method to divide photovoltaic components according to the predetermined classification standard to form different regional matrices;

[0174] Component classification module: classifies the photovoltaic components into corresponding regional matrices according to the division results.

[0175] The working principle of the above technical solution is: extracting predetermined classification standards, which can be parameters such as size, material, color, thickness, etc. of photovoltaic modules, or indicators such as electrical performance, optical performance, mechanical performance, etc.; using matrix analysis method, taking the extracted classification standards as dimensions, performing data analysis and matrix operations on different dimensions, and obtaining different regional matrices. These regional matrices can be composed of multiple attributes to form a multidimensional attribute space, and different regional divisions can be obtained by combining different attributes; according to the classification results, photovoltaic modules are classified into corresponding regional matrices according to their characteristics and attributes to realize the classification management and refined management of photovoltaic modules.

[0176] The effect of the above technical solution is: by dividing the photovoltaic modules according to the predetermined classification standards, different types of photovoltaic modules can be managed in a refined manner. In this way, the performance characteristics and operating status of different types of photovoltaic modules can be better understood and mastered, and management strategies and maintenance plans can be formulated in a targeted manner; by dividing the photovoltaic modules into different regional matrices, more accurate data analysis and management can be achieved. According to the characteristics and properties of the photovoltaic modules in different regional matrices, targeted optimization and adjustment can be carried out to improve the power generation efficiency and performance of the photovoltaic modules; refined management and optimization adjustment can help identify and deal with problems and faults in photovoltaic modules. By timely discovering and solving problems, the cost of maintenance and replacement can be reduced, the service life of photovoltaic modules can be extended, and thus the operating cost can be reduced; by dividing the photovoltaic modules into regional matrices, the quality status of photovoltaic modules can be better grasped. For photovoltaic modules in different regional matrices, targeted quality control and monitoring measures can be taken to improve the overall quality level of photovoltaic modules; by analyzing the data of photovoltaic modules in the regional matrix, more accurate and comprehensive information can be obtained. This information can provide reference and support for decision makers, help them make scientific and reasonable operation and management decisions, and improve the overall benefits of photovoltaic power stations.

[0177] Embodiment 9,

[0178] In this embodiment, the parameter determination module includes:

[0179] Retrieving the cleaning requirements of the photovoltaic power station;

[0180] Demand input module: inputs the cleaning demand into the trained and optimized deep learning model;

[0181] Demand determination module: inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model;

[0182] Parameter determination module: for the strings that need to be cleaned, the cleaning parameters are determined according to the operating parameters of the strings using the trained and optimized deep learning model, wherein the cleaning parameters include cleaning sequence, cleaning intensity and cleaning duration;

[0183] Cleaning scheduling module: according to the cleaning intensity and cleaning time of each string to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning of the strings to be cleaned is scheduled according to the cleaning order.

[0184] The working principle of the above technical solution is: by retrieving the cleaning requirements of the photovoltaic power station, it is input into the deep learning model that has been trained and optimized. Then, the operating parameters of each string of the photovoltaic power station are input into the deep learning model, and the deep learning model is used to analyze and judge the input operating parameters to determine whether each string needs to be cleaned. For the strings that need to be cleaned, the deep learning model is used to determine the cleaning parameters according to their operating parameters, including the cleaning order, cleaning intensity and cleaning time. Finally, according to the cleaning intensity and cleaning time of each string that needs to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning schedule of the strings to be cleaned is carried out according to the determined cleaning order.

[0185] The effect of the above technical solution is: by completing the training and optimization of the deep learning model, it is possible to accurately determine whether each group of strings in the photovoltaic power station meets the cleaning requirements. Based on the analysis and judgment of the deep learning model, the error of subjective judgment can be eliminated and the accuracy of the cleaning judgment can be improved; using the deep learning model, the entire cleaning process can be automatically executed, reducing the need for manual operation. Only by inputting the cleaning requirements of the photovoltaic power station and the operating parameters of each group of strings, the deep learning model can automatically judge and determine the cleaning parameters, greatly improving the efficiency and convenience of the cleaning process; by completing the training and optimization of the deep learning model, the cleaning parameters can be determined according to the operating parameters of the group of strings, including the cleaning sequence, cleaning intensity and cleaning duration. In this way, personalized cleaning treatment can be carried out for different groups of strings, maximizing the cleaning effect and ensuring the normal operation and power generation efficiency of the photovoltaic power station; due to the precise determination of the cleaning parameters, the cleaning intensity and duration can be allocated according to actual needs, avoiding unnecessary cleaning waste and resource consumption, thereby saving cleaning costs. At the same time, automated cleaning scheduling can optimize the use of cleaning time and human resources, further reducing the cost of cleaning.

[0186] Embodiment 10

[0187] In this embodiment, the effect evaluation module includes:

[0188] Real-time monitoring module: real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree;

[0189] Degree evaluation module: evaluates the cleaning degree in combination with the regional matrix information;

[0190] State judgment module: evaluate the power generation efficiency of the photovoltaic module using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data, obtain the evaluation results, and judge whether the photovoltaic module is in the best performance state through the evaluation results.

[0191] The working principle of the above technical solution is: real-time monitoring of data parameters during string cleaning, including performance parameters and cleaning degree. These parameters can be monitored and collected in real time by sensors and other equipment to reflect the actual situation of the string. The cleaning degree is evaluated in combination with regional matrix information. Regional matrix is ​​a commonly used evaluation method in photovoltaic power stations. The effect of string cleaning can be obtained by comparing the differences in power generation efficiency of components in different regions. Using regional matrix information, the cleaning degree of the string can be objectively evaluated. The power generation efficiency of photovoltaic components is evaluated by using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data. The deep learning model can predict the power generation efficiency of the component through historical data and real-time monitoring data, and compare it with the historical optimal power generation efficiency to obtain the current power generation efficiency of the component. According to the evaluation results, it is judged whether the photovoltaic component is in the best performance state. According to the real-time monitoring data and evaluation results, it can be judged whether the component has reached the best performance state. If not, further cleaning or other treatment is required.

[0192] The effects of the above technical solution are: by real-time monitoring of the performance parameters and cleaning degree during the string cleaning process, the operation status of the string can be obtained in time to ensure accurate evaluation and maintenance of the photovoltaic modules; combined with the regional matrix information, the cleaning degree can be evaluated. In this way, the cleaning effect of the string can be judged more objectively to avoid performance degradation or damage caused by incomplete or excessive cleaning; the power generation efficiency of the photovoltaic module is evaluated using the prediction results of the matrix area cleanliness of the deep learning model and the historical optimal power generation efficiency data. By comparing with the historical optimal power generation efficiency, the current power generation efficiency of the module can be understood and abnormal conditions can be discovered in time; through the evaluation results, it can be determined whether the photovoltaic module is in the best performance state. If the evaluation results show that the module is not in the best state, corresponding measures can be taken, such as further cleaning, repair or replacement of the module, to ensure the normal operation and power generation efficiency of the photovoltaic power station.

[0193] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A photovoltaic power station cleaning process evaluation method based on matrix analysis, characterized in that: The photovoltaic power station cleaning process evaluation method comprises: Collecting the operating parameter information of the photovoltaic power station in real time and sending the operating parameter information to the central system; Using a predetermined classification standard, photovoltaic modules are divided into different regional matrices; The deep learning model is trained using the regional matrix information of the regional matrix combined with the historical performance data of the photovoltaic power station to obtain a trained and optimized deep learning model; Determine the photovoltaic power station strings that need to be cleaned and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements; Evaluate the cleaning effect based on the evaluation results of the strings; The real-time collection of operating parameter information of the photovoltaic power station and sending the operating parameter information to the central system include: Controlling the data acquisition device to periodically measure the current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station; Sending the operating parameter information to the central system via the communication module; According to the cleaning requirements, the strings of the photovoltaic power station that need to be cleaned and their cleaning parameters are determined by the trained and optimized deep learning model, including: Retrieving the cleaning requirements of the photovoltaic power station; Inputting the cleaning requirements into the trained and optimized deep learning model; Inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model; For the strings that need to be cleaned, the trained and optimized deep learning model is used to determine the cleaning parameters according to the operating parameters of the strings, wherein the cleaning parameters include the cleaning sequence, cleaning intensity and cleaning duration; The specific strength ratio of the cleaning strength is obtained by the following formula: in, P A specific intensity ratio indicating the intensity of cleaning; P 0 represents the preset ratio reference value, i.e. the initialization ratio; M represents the dust parameter matrix; M 0 represents the benchmark parameter matrix; m h and m hs Respectively represent the dust mass and dust distribution area of ​​the component during the current cleaning; Δ m h and Δ m hs They respectively represent the mass change of the component dust and the dust distribution area change between the current cleaning and the last cleaning; m hp and m hps Respectively represent the preset component dust mass reference value and dust distribution area reference value corresponding to the ratio reference value; Furthermore, the cleaning strength is obtained by using the specific strength ratio of the cleaning strength, wherein the cleaning strength is obtained by the following formula: in, F Indicates the intensity of cleaning; F 0 represents the preset initial value of cleaning intensity; At the same time, the cleaning time is obtained by the following formula: in, T Indicates the cleaning time; T 0 indicates the preset initial value of cleaning time; According to the cleaning intensity and cleaning time of each string to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning of the strings to be cleaned is scheduled according to the cleaning sequence.

2. The photovoltaic power station cleaning process evaluation method based on matrix analysis according to claim 1 is characterized in that: PV panels are divided into different regional matrices using pre-defined classification criteria, including: extracting the predetermined classification criteria; Using a matrix analysis method to divide the photovoltaic modules according to the predetermined classification standard to form different regional matrices; The photovoltaic components are classified into corresponding regional matrices according to the division results.

3. The photovoltaic power station cleaning process evaluation method based on matrix analysis according to claim 1 is characterized in that: The cleaning effect is evaluated based on the evaluation results of the string, including: Real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree; evaluating the cleaning degree in combination with the regional matrix information; The power generation efficiency of the photovoltaic module is evaluated using the prediction results of the matrix area cleanliness of the trained and optimized deep learning model and the historical optimal power generation efficiency data to obtain an evaluation result, and the evaluation result is used to determine whether the photovoltaic module is in the best performance state.

4. A photovoltaic power station cleaning process evaluation system based on matrix analysis, characterized in that: The photovoltaic power station cleaning process evaluation system comprises: Information collection module: collects the operating parameter information of the photovoltaic power station in real time and sends the operating parameter information to the central system; Matrix division module: divides photovoltaic modules into different regional matrices using predetermined classification standards; Model acquisition module: using the regional matrix information of the regional matrix combined with the historical performance data of the photovoltaic power station to train the deep learning model, and obtain a trained and optimized deep learning model; Parameter determination module: determining the strings that need to be cleaned in the photovoltaic power station and their cleaning parameters through the trained and optimized deep learning model according to the cleaning requirements; Effect evaluation module: Evaluate the cleaning effect based on the evaluation results of the string; Wherein, the information collection module includes: Parameter measurement module: control the data acquisition device to regularly measure the current, voltage, temperature and light intensity of each string included in the photovoltaic power station; wherein the current, voltage, temperature and light intensity of each string are the operating parameter information of the photovoltaic power station; Information transmission module: sending the operating parameter information to the central system through the communication module; Wherein, the parameter determination module includes: Retrieving the cleaning requirements of the photovoltaic power station; Demand input module: inputs the cleaning demand into the trained and optimized deep learning model; Demand determination module: inputting the operating parameters of each group of strings of the photovoltaic power station into the trained and optimized deep learning model, and determining whether each group of strings of the photovoltaic power station meets the cleaning requirements through the trained and optimized deep learning model; Parameter determination module: for the strings that need to be cleaned, the cleaning parameters are determined according to the operating parameters of the strings using the trained and optimized deep learning model, wherein the cleaning parameters include cleaning sequence, cleaning intensity and cleaning duration; The specific strength ratio of the cleaning strength is obtained by the following formula: in, P A specific intensity ratio indicating the intensity of cleaning; P 0 represents the preset ratio reference value, i.e. the initialization ratio; M represents the dust parameter matrix; M 0 represents the benchmark parameter matrix; m h and m hs Respectively represent the dust mass and dust distribution area of ​​the component during the current cleaning; Δ m h and Δ m hs They respectively represent the mass change of the component dust and the dust distribution area change between the current cleaning and the last cleaning; m hp and m hps Respectively represent the preset component dust mass reference value and dust distribution area reference value corresponding to the ratio reference value; Furthermore, the cleaning strength is obtained by using the specific strength ratio of the cleaning strength, wherein the cleaning strength is obtained by the following formula: in, F Indicates the intensity of cleaning; F 0 represents the preset initial value of cleaning intensity; At the same time, the cleaning time is obtained by the following formula: in, T Indicates the cleaning time; T 0 indicates the preset initial value of cleaning time; Cleaning scheduling module: according to the cleaning intensity and cleaning time of each string to be cleaned, the cleaning intensity and cleaning time are allocated, and the cleaning of the strings to be cleaned is scheduled according to the cleaning order.

5. The photovoltaic power station cleaning process evaluation system based on matrix analysis according to claim 4 is characterized in that: The matrix partitioning module comprises: Standard extraction module: extracting the predetermined classification standard; Component division module: using a matrix analysis method to divide photovoltaic components according to the predetermined classification standard to form different regional matrices; Component classification module: classifies the photovoltaic components into their corresponding regional matrices according to the division results.

6. The photovoltaic power station cleaning process evaluation system based on matrix analysis according to claim 4 is characterized in that: The effect evaluation module comprises: Real-time monitoring module: real-time monitoring of data parameters during string cleaning, wherein the data parameters include performance parameters and cleaning degree; Degree evaluation module: evaluates the cleaning degree in combination with the regional matrix information; State judgment module: evaluate the power generation efficiency of the photovoltaic module by using the prediction results of the matrix area cleanliness of the trained and optimized deep learning model and the historical optimal power generation efficiency data, obtain the evaluation results, and judge whether the photovoltaic module is in the best performance state through the evaluation results.

Images