Method and system for evaluating running state of photovoltaic power generation system

Abstract

The invention discloses a photovoltaic power generation system operation state evaluation method and system, and the method comprises the steps: carrying out the subjective empowerment of an evaluation index of a photovoltaic power generation system to be evaluated through employing a subjective empowerment method, carrying out the objective empowerment of the evaluation index through employing a plurality of objective empowerment methods, carrying out the combined empowerment of a subjective weight and an objective weight, and obtaining a combined weight, the evaluation indexes and the combination weights form feature vectors, the feature vectors are input into a trained support vector machine model for operation state evaluation, an operation state evaluation result is obtained, the subjective and objective combination weighting method effectively fuses subjective experience and objective data driven weight information, the relative importance of all operation parameters is accurately quantified, and the operation state evaluation accuracy is improved. The support vector machine model has remarkable advantages when processing small samples and high-dimensional data by virtue of the strong nonlinear classification and generalization capability, the operation state can be efficiently and accurately judged according to multi-parameter characteristics, and the operation state of the photovoltaic power generation system can be efficiently and accurately evaluated by combining subjective and objective combination weighting and the support vector machine model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation state evaluation, in particular to a photovoltaic power generation system operation state evaluation method and system. BACKGROUND

[0002] With the growing demand for clean energy worldwide, distributed photovoltaic power generation, as an important form of renewable energy utilization, is being widely popularized and applied. With its flexible installation method, on-site power consumption and other advantages, it has shown great potential in many fields such as residential, commercial buildings and small industrial facilities, and has made significant contributions to alleviating energy crisis and reducing carbon emissions.

[0003] However, the operation of distributed photovoltaic power generation system is influenced by a variety of complex factors. On the one hand, natural environmental factors such as the intermittency of light intensity and the large fluctuation of temperature constantly affect the power generation efficiency of photovoltaic components; on the other hand, the aging of equipment itself, the performance stability of inverter and the adaptability of grid connection and other technical factors also play a key role in the overall operation state of the system. The dynamic changes of these factors make it a challenging task to accurately and timely grasp the operation state of the distributed photovoltaic power generation system.

[0004] Traditional operation state monitoring methods often focus on the analysis of a single indicator or a few parameters, and it is difficult to comprehensively consider the deep correlation behind multi-source and heterogeneous data, resulting in problems such as untimely warning of complex faults and inaccurate state evaluation. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a photovoltaic power generation system operation state evaluation method and system, which can efficiently and accurately evaluate the operation state of the photovoltaic power generation system.

[0006] To solve the above technical problems, the technical scheme adopted by the present application is: A photovoltaic power generation system operation state evaluation method, comprising the steps of: obtaining evaluation indexes of a photovoltaic power generation system to be evaluated; using a subjective weighting method to subjectively weight the evaluation indexes to obtain subjective weights; using a variety of objective weighting methods to objectively weight the evaluation indexes to obtain objective weights; combining the subjective weights and the objective weights to obtain combined weights; inputting the evaluation indexes and the combined weights into a trained support vector machine model to perform operation state evaluation, and obtaining the operation state evaluation result of the photovoltaic power generation system to be evaluated.

[0007] To solve the above technical problems, another technical solution adopted by the present application is: A photovoltaic power generation system operation state evaluation system, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements each step of the above-mentioned photovoltaic power generation system operation state evaluation method when executing the computer program.

[0008] The present application has the beneficial effects that: the evaluation indexes of the photovoltaic power generation system to be evaluated are obtained, the subjective weighting method is used for subjective weighting of the evaluation indexes to obtain subjective weights, a plurality of objective weighting methods are used for objective weighting of the evaluation indexes to obtain objective weights, the subjective weights and the objective weights are combined to obtain combined weights, the evaluation indexes and the combined weights are input into the trained support vector machine model to perform operation state evaluation to obtain operation state evaluation results, the subjective and objective combined weighting method effectively fuses the subjective experience and the objective data-driven weight information, accurately quantifies the relative importance of each operation parameter, and the support vector machine model has strong nonlinear classification and generalization ability, has obvious advantages in processing small samples and high-dimensional data, and can accurately distinguish the operation state according to multiple parameter characteristics, so that the combination of the subjective and objective combined weighting and the support vector machine model can efficiently and accurately evaluate the operation state of the photovoltaic power generation system. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 A flowchart of a photovoltaic power generation system operation state evaluation method according to an embodiment of the present application; Figure 2 A schematic diagram of a photovoltaic power generation system operation state evaluation system according to an embodiment of the present application; Figure 3 A comparison diagram of actual operation state and predicted operation state in a photovoltaic power generation system operation state evaluation method according to an embodiment of the present application; Figure 4 An SVM classification result diagram and a support vector diagram in a photovoltaic power generation system operation state evaluation method according to an embodiment of the present application. DETAILED DESCRIPTION

[0010] To explain the technical content, the achieved purposes and effects of the present application in detail, the following will be explained in combination with the embodiments and the accompanying drawings.

[0011] Before the embodiments of the present application are described in detail, some related concepts will be explained first: Support vector machine (SVM) model: a generalized linear classifier that classifies data in a binary manner according to a supervised learning method, and the decision boundary is a maximum margin hyperplane solved for the learning samples; Radial Basis Function (RBF kernel): a kind of radial symmetry scalar function; Three-scale analytic hierarchy process: an improved decision-making method based on analytic hierarchy process (AHP), which simplifies the construction of judgment matrix and converts traditional nine-scale comparison into three scales (0, 1, 2), reducing the complexity of subjective judgment; Mean square deviation weighting method: an objective weighting method based on data difference, which determines the weight by calculating the deviation degree of data from the mean value; Grey correlation degree weighting method: a method for determining weights in multi-attribute decision analysis, which captures the non-linear relationship between indicators by calculating the geometric similarity of each indicator to the ideal optimal value sequence.

[0012] In the prior art, the evaluation of the operating state of the photovoltaic power generation system often focuses on the analysis of a single indicator or a few parameters, and it is difficult to comprehensively consider the deep-seated correlation behind multi-source and heterogeneous data, resulting in problems such as untimely warning of complex faults and inaccurate state evaluation.

[0013] To at least solve the above problems, please refer to Figure 1 The embodiment of the present application provides a photovoltaic power generation system operating state evaluation method, comprising the steps of: obtaining evaluation indicators of a photovoltaic power generation system to be evaluated; subjectively weighting the evaluation indicators using a subjective weighting method to obtain subjective weights; objectively weighting the evaluation indicators using a plurality of objective weighting methods to obtain objective weights; combining the subjective weights and the objective weights to obtain combined weights; inputting the evaluation indicators and the combined weights into a trained support vector machine model to perform operating state evaluation, and obtaining the operating state evaluation result of the photovoltaic power generation system to be evaluated.

[0014] Further, before obtaining the evaluation indicators of the photovoltaic power generation system to be evaluated, the method further comprises: defining the radial basis kernel function of the support vector machine model; obtaining a training data set; training the support vector machine model using the training data set according to the radial basis kernel function, solving the model optimization problem using a sequential minimal optimization algorithm, obtaining the parameters of the trained support vector machine model, and thus completing the training of the support vector machine model.

[0015] Further, the radial basis kernel function of the support vector machine model is defined as: ; In the formula, K ( x i , x j ) represents a radial basis kernel function, x i represents a feature vector i , x j represents a feature vector j , represents a kernel parameter.

[0016] From the above description, the radial basis kernel function is selected, the support vector machine model is trained using the training data set according to the radial basis kernel function, and the model optimization problem is solved using the sequential minimal optimization algorithm, which can finely tune the model parameters and effectively improve the adaptability and reliability of the support vector machine model.

[0017] Further, the evaluation indexes are subjectively weighted using a subjective weighting method to obtain subjective weights, including: using a three-scale analytic hierarchy process to construct a comparison matrix of the evaluation indexes; calculating a judgment matrix according to the comparison matrix; calculating the subjective weights of the evaluation indexes according to the judgment matrix.

[0018] Further, the three-scale analytic hierarchy process is used to construct the comparison matrix of the evaluation indexes, specifically: ; ; In the formula, A represents a comparison matrix, n represents the number of evaluation indexes, a ij represents the comparison matrix element of index i and index j ; calculating a judgment matrix according to the comparison matrix, specifically: ; ; In the formula, D represents a judgment matrix, d ij represents the judgment matrix element of index i and index j , a ik represents the comparison matrix element of index i and index k , akj Indicators k With indicators j Compare matrix elements; The subjective weights of the evaluation indicators are calculated based on the judgment matrix, specifically as follows: ; In the formula, a j Indicators j Subjective weighting.

[0019] As described above, the three-scale analytic hierarchy process (AHP) first constructs a comparison matrix of evaluation indicators, then calculates the judgment matrix and subjective weights, which can improve the credibility of the weights. At the same time, it takes into account both qualitative judgment and quantitative calculation, making it easier to quickly capture core indicators in complex evaluation systems and improve evaluation efficiency and the persuasiveness of the results.

[0020] Furthermore, the evaluation indicators are objectively weighted using various objective weighting methods, resulting in objective weights including: The evaluation indicators are objectively weighted using the mean square error weighting method to obtain the first objective weight; The evaluation indicators are objectively weighted using the grey relational analysis method to obtain the second objective weight. The objective weight is obtained by combining the first objective weight and the second objective weight.

[0021] As described above, combining the mean square error weighting method and the grey relational degree weighting method to obtain objective weights can complement the data fluctuation and shape information, avoid the distortion of a single method, enhance the indicator discrimination, retain the dynamic correlation characteristics, make the weights both robust and sensitive, and improve the interpretability and reliability of the evaluation results.

[0022] Furthermore, the evaluation indicators are objectively weighted using the mean squared error weighting method, resulting in the first objective weights, which include: Calculate the mean square error of the evaluation index; The mean squared error is normalized to obtain the first objective weight.

[0023] As described above, the mean square error weighting method assigns weights based on the difference between the evaluation object and the mean, reflecting the comparative strength of the indicator data.

[0024] Furthermore, the evaluation indicators are objectively weighted using the grey relational analysis method, resulting in a second set of objective weights, including: The maximum value of each column in the evaluation index is selected as the positive ideal solution, and the minimum value of each column in the evaluation index is selected as the negative ideal solution. calculating a first distance of the evaluation index from the positive ideal solution and a second distance of the evaluation index from the negative ideal solution using a Mahalanobis distance; calculating a closeness of the evaluation index from the positive ideal solution according to the first distance and the second distance; calculating a group grey correlation degree of the evaluation index according to the closeness; normalizing the group grey correlation degree to obtain a second objective weight.

[0025] As can be seen from the above description, the grey correlation degree weighting method weights according to the difference between the evaluation index and the optimal value and the worst value, and weights through the fluctuation degree of the index data, which is more accurate.

[0026] Further, the subjective weight and the objective weight are combined to obtain a combined weight, including: constructing a combined weighting model based on the subjective weight and the objective weight; solving the subjective weight coefficient and the objective weight coefficient in the combined weighting model by using a Lagrange extreme value method; combining the subjective weight and the objective weight according to the subjective weight coefficient and the objective weight coefficient to obtain a combined weight.

[0027] As can be seen from the above description, the combined weight accurately reflecting the importance of each parameter can be obtained through linear combination.

[0028] Please refer to Figure 2 Another embodiment of the present application provides a photovoltaic power generation system operation state evaluation system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and each step of the above-mentioned photovoltaic power generation system operation state evaluation method is realized when the processor executes the computer program.

[0029] The above-mentioned photovoltaic power generation system operation state evaluation method and system of the present application are suitable for distributed photovoltaic power generation system operation state evaluation scenarios, and the following will be described through specific embodiments: Please refer to Figure 1 An embodiment of the present application is: A photovoltaic power generation system operation state evaluation method, comprising the steps of: S1, defining a radial basis kernel function of a support vector machine model, specifically: ; In the formula, K ( x i , x j ) represents the radial basis kernel function, xi Representing the eigenvector i , x j Representing the eigenvector j , Indicates kernel parameters.

[0030] S2. Obtain the training dataset.

[0031] The training dataset is {( x 1, y 1),( x 2, y 2),…,( x l , y l )}, x i It is the first i The feature vector of each sample (including preprocessed evaluation index parameters and corresponding weights). y i It is the category label of the sample (such as the running status being divided into normal, abnormal-1, abnormal-2, etc.).

[0032] The parameters of the SVM model are determined using cross-validation: for the RBF kernel function, the penalty parameter needs to be determined. C and kernel parameters The dataset is divided into training, validation, and test sets, typically in a certain ratio (e.g., 7:2:1). An SVM model is trained on the training set, and parameters are adjusted on the validation set to achieve the best classification performance. For each combination (…), C , ), calculate the model's performance metrics on the validation set, and select the parameter combination with the best performance.

[0033] S3. Train the support vector machine model using the training dataset according to the radial basis kernel function, solve the model optimization problem using the sequence minimum optimization algorithm, and obtain the parameters of the trained support vector machine model, thereby completing the training of the support vector machine model.

[0034] Specifically, the model optimization problem is as follows: ; In the formula, w Describes the normal vector of the hyperplane. Represents slack variables. b Indicates the bias term. ( x i ) indicates that the input vector x ia function mapping to a high-dimensional feature space (determined by a kernel function); The model optimization problem is solved using a sequential minimal optimization algorithm, and the model parameters w and b .

[0035] S4, obtaining an evaluation index of the photovoltaic power generation system to be evaluated.

[0036] The evaluation index can be divided into four types: safety quality management index, operation and maintenance management index, operation and maintenance performance index, and production operation index. The safety quality management index includes the number of minor injuries, the number of fires, and the number of inverter and transformer equipment damage and non-repairable times. The operation and maintenance management index includes the completeness rate of equipment repair and maintenance history records, the completeness rate of equipment failure and accident history records, and the completeness rate of spare parts and spare parts warehouse records. The operation and maintenance performance index includes the plan completion rate, equipment failure loss equivalent hours, and power station abnormal loss equivalent hours. The production operation index includes the field station energy consumption index and the comprehensive plant power consumption rate.

[0037] The number of minor injuries: the number of injuries of operation and maintenance personnel in the photovoltaic power station during the statistical period, which represents the ability of the power station to ensure personnel safety, specifically: ; In the formula, R W represents the number of minor injuries, W k represents the number of injuries of operation and maintenance personnel in the photovoltaic power station on the k th day ( k =1, 2, 3,.. m ) during the statistical period.

[0038] The number of fires: the number of fires in the photovoltaic power station or within 100 meters of the power station during the statistical period, which represents the management quality of the power station for fire safety, specifically: ; In the formula, R V represents the number of fires, V k represents the number of fires in the photovoltaic power station or within 100 meters of the power station on the k th day during the statistical period.

[0039] The number of inverter and transformer equipment damage and non-repairable times: the number of major damage of inverters and transformers in the photovoltaic power station during the statistical period, which represents the ability of the power station to ensure equipment safety, specifically: ; In the formula, RU The number of times that inverter and transformer equipment damage occurs and is not repairable, U k The number of times that inverter and transformer equipment damage occurs and is not repairable, k The number of times that inverter and transformer equipment damage occurs and is not repairable,

[0040] Equipment repair and maintenance history record completeness: the ratio of the number of times that the operation and maintenance personnel of the photovoltaic power station complete records to the number of times of equipment repair and maintenance during the statistical period, indicating the completeness of the operation and maintenance personnel in recording the repair and maintenance process content, specifically: ; In the formula, R OR The number of times that inverter and transformer equipment damage occurs and is not repairable, C F The number of times that the operation and maintenance personnel of the photovoltaic power station complete records, C D The number of times of equipment repair and maintenance of the photovoltaic power station.

[0041] Equipment failure and accident history record completeness: the ratio of the number of times that the operation and maintenance personnel of the photovoltaic power station complete records to the number of times of equipment failure and accident during the statistical period, indicating the completeness of the operation and maintenance personnel in recording the failure content, specifically: ; In the formula, R AR The number of times that inverter and transformer equipment damage occurs and is not repairable, C H The number of times of equipment failure and accident of the photovoltaic power station.

[0042] The number of times that inverter and transformer equipment damage occurs and is not repairable, ; In the formula, R SP The number of times that inverter and transformer equipment damage occurs and is not repairable, C IO The number of times of equipment failure and accident of the photovoltaic power station.

[0043] The number of times that inverter and transformer equipment damage occurs and is not repairable, ; In the formula, R PF The number of times that inverter and transformer equipment damage occurs and is not repairable, E RPinv represents the actual power generation of the inverter, in kWh, E P Pplan represents the planned power generation, in kWh.

[0044] Equipment failure loss equivalent hours: the ratio of the loss of power due to equipment failure in the photovoltaic power station to the installed capacity during the statistical period, indicating how much power is lost due to equipment failure in the photovoltaic power station, specifically: ; wherein, Y DF Pinv represents the actual power generation of the inverter, in kWh, E F Ploss represents the loss of power due to equipment failure in the photovoltaic power station, in kWh, P 0 represents the installed capacity of the power station, in kWh.

[0045] Station abnormal loss equivalent hours: the ratio of the loss of power due to abnormality in the photovoltaic power station to the installed capacity during the statistical period, indicating how much power is lost due to abnormality in the photovoltaic power station, specifically: ; wherein, Y AL Pinv represents the actual power generation of the inverter, in kWh, E A Ploss represents the loss of power due to equipment failure in the photovoltaic power station, in kWh, P 0 represents the installed capacity of the power station, in kWh.

[0046] Station energy consumption index: the ratio of the power output to the inverter power generation during the statistical period, indicating the degree of power consumption from the grid by the station, specifically: ; wherein, R Pinv represents the actual power generation of the inverter, in kWh, E in Pout represents the power output of the photovoltaic power station, in kWh, E P Pinv represents the actual power generation of the inverter, in kWh.

[0047] Comprehensive plant power consumption rate: the percentage of the comprehensive plant power consumption to the power generation of the photovoltaic power station during the statistical period, wherein the comprehensive plant power consumption refers to the total power consumption during the production and operation of the photovoltaic power station, including the power consumption of the power generation unit, the transformer, the power collection line, the electrical equipment in the booster station (including the main transformer, the station transformer loss, and the busbar, etc.), and the transmission line, etc., specifically: ; wherein, Z Pinv represents the actual power generation of the inverter, in kWh,E TC represents the comprehensive plant power consumption, and the unit is kWh.

[0048] In an alternative embodiment, S5 is further preceded by: data preprocessing is performed on the evaluation indexes to obtain preprocessed evaluation indexes.

[0049] The data preprocessing includes sequentially performing elimination of null values, standardization, and normalization.

[0050] For example, the index data of m power stations are collected for data preprocessing, and null values are first eliminated to obtain a data set containing n indexes X = ( x 1, x i , x n ), and then the data set is standardized to obtain a standardized index data set , specifically: ; In the formula, represents the mean of the data set, represents the standard deviation of the data set; Finally, each index is normalized to obtain a preprocessed evaluation index data set Y = ( y 1, y i , y n ), specifically: ; In the formula, represents the minimum value in the data set, represents the maximum value in the data set.

[0051] The subsequent steps can process the preprocessed evaluation indexes.

[0052] S5, subjective weighting is performed on the evaluation indexes using a subjective weighting method to obtain subjective weights, specifically including S51-S53: S51, a three-scale analytic hierarchy process is used to construct a comparison matrix of the evaluation indexes, specifically: ; ; In the formula, A represents the comparison matrix, n represents the number of evaluation indexes, aij Indicators i With indicators j Compare matrix elements.

[0053] S52. Calculate the judgment matrix based on the comparison matrix, specifically as follows: ; ; In the formula, D Represents the judgment matrix. d ij Indicators i With indicators j Determine the elements of the matrix. a ik Indicators i With indicators k Comparison of matrix elements, a kj Indicators k With indicators j Compare matrix elements.

[0054] S53. Calculate the subjective weights of the evaluation indicators based on the judgment matrix, specifically as follows: ; In the formula, a j Indicators j Subjective weighting.

[0055] S6. Apply multiple objective weighting methods to the evaluation indicators to obtain objective weights, specifically including S61-S63: S61. The evaluation indicators are objectively weighted using the mean square error weighting method to obtain the first objective weight, specifically including S611-S612: S611. Calculate the root mean square error of the evaluation index, specifically as follows: ; In the formula, σ j Indicators j The mean squared error, x ij Indicators j The next i One data sample, Indicators j The mean.

[0056] S612. Normalize the mean squared error to obtain the first objective weight, specifically as follows: ; In the formula,w 3j Indicators j The first objective weight.

[0057] The mean square error weighting method assigns weights based on the difference between the evaluated object and the mean, reflecting the comparative strength of the indicator data.

[0058] S62. The evaluation indicators are objectively weighted using the grey relational degree weighting method to obtain the second objective weight, specifically including S621-S625.

[0059] S621. Select the maximum value of each column in the evaluation index as the positive ideal solution, and select the minimum value of each column in the evaluation index as the negative ideal solution.

[0060] The positive ideal solution can be expressed as: The negative ideal solution can be expressed as: .

[0061] S622. Calculate the first distance between the evaluation index and the positive ideal solution and the second distance between the evaluation index and the negative ideal solution using Mahalanobis distance, specifically as follows: ; ; In the formula, Indicates the first distance. Indicates the second distance. express m Evaluation metrics for each sample k The vector formed by them P -1 This represents the inverse of the sample's covariance matrix.

[0062] S623. Calculate the closeness of the evaluation index to the positive ideal solution based on the first distance and the second distance, specifically as follows: ; In the formula, Indicators k The degree of closeness to the ideal solution.

[0063] S624. Calculate the group grey relational degree of the evaluation index based on the proximity, specifically as follows: ; In the formula, Indicators j The gray relational degree of the group.

[0064] S625. Normalize the group gray relational degree to obtain the second objective weight, specifically: ; In the formula, w 4j This indicates the second objective weight.

[0065] The grey relational weighting method assigns weights based on the difference between the evaluation index and the optimal and worst values, and also assigns weights based on the degree of fluctuation of the index data.

[0066] S63. Combine the first objective weight and the second objective weight to obtain the objective weight.

[0067] Specifically, the prior probability is known to be q ( x Assume the objective weights of the unknown probability combinations are... p ( x The probability distribution constraints are as follows: p ( x ) dx =1, and the best estimate of the true probability is: ; The first and second objective weights are prior probabilities, and need to be combined. =( ,…, The distances between them are extremely close, specifically: ; ; The Lagrange function is constructed to derive the formula for calculating the combined weights. After normalization, the final objective weights are obtained, as follows: ; In the formula, Indicators j Objective weighting.

[0068] S7. Combine and assign weights to the subjective weights and the objective weights to obtain combined weights, specifically including S71-S73: S71. Construct a combined weighting model based on the subjective weights and the objective weights.

[0069] Specifically, calculate the first j One photovoltaic power generation system to be evaluated and the average vector Weighted distance d j Specifically: ; ; In the formula, v iIndicates the first i The coefficient of variation of each evaluation indicator Indicates the first i The combined weights of the evaluation indicators y ij Indicates the first i The first evaluation indicator j One observation (or sample). Indicates the first i The average of each evaluation indicator, Indicates the first i Standard deviation of each evaluation indicator; coefficient of variation v i The larger the value, the more likely it is to be the first. i The greater the variability in the distribution of each indicator in the comprehensive evaluation and the greater its information content, the more likely the combined weighting model is: ; In the formula, k 1 represents the subjective weighting coefficient. k 2 represents the objective weighting coefficient. Indicates subjective weighting. Indicates objective weight.

[0070] S72. The subjective weight coefficients and objective weight coefficients in the combined weighting model are solved using the Lagrange extreme value method, specifically as follows: ; right k 1. k 2. Perform normalization to obtain the normalized subjective weight coefficients and objective weight coefficients. k 1' k 2', specifically: .

[0071] S73. The subjective weight and the objective weight are combined and weighted according to the subjective weight coefficient and the objective weight coefficient to obtain the combined weight, specifically as follows: .

[0072] The maximum coefficient of variation method is used to combine and weight subjective and objective factors, constructing an objective function that maximizes the combined weights after coefficient of variation weighting, thereby determining the combination coefficients of subjective and objective factors. k 1. k 2. This reflects the idea that the greater the variability in the distribution of subjective and objective weights in combined weighting, the greater the weight.

[0073] In one alternative implementation, it further includes: An evaluation score of the evaluation index is calculated according to the evaluation index and the combination weight SCORE , specifically: .

[0074] For example, after obtaining the weight of the photovoltaic power station operation and maintenance quality evaluation index, the standardized power station data is weighted to obtain the comprehensive evaluation score of the power station operation and maintenance quality.

[0075] S8, the evaluation index and the combination weight are input into the trained support vector machine model to perform running state evaluation, and an operation state evaluation result of the photovoltaic power generation system to be evaluated is obtained.

[0076] The operation state evaluation result includes two results of normal or abnormal, such as output of "1 normal" indicating normal system operation, and output of "0 abnormal" indicating abnormal power output.

[0077] As shown in Figure 3 , Figure 3 , a comparison between the actual state and the model predicted state is shown. Specifically, the circle point represents the actual state (i.e. the true label of the test data), where 0 represents the normal state and 1 represents the abnormal state. The dot represents the predicted state (i.e. the label predicted by the SVM model), and again, 0 represents normal and 1 represents abnormal. The X-axis represents the serial number of each sample in the test set. The Y-axis is the running state (0 or 1). It can be seen intuitively that the difference between the prediction result of the model and the actual state: if the predicted state coincides with the actual state, it means that the sample is correctly classified. If the predicted state is different from the actual state, it means that the sample is misclassified. When the dot coincides with the circle point completely, it means that the prediction of the sample is correct. When the dot does not coincide with the circle point, it means that the sample is misclassified. It can be seen that Figure 3 most of the dots should coincide with the circle points, indicating that the prediction accuracy of the model is high.

[0078] As shown in Figure 4 , Figure 4 , the classification result of the SVM model and the distribution of support vectors are shown. Specifically, the × point and the small circle point represent the SVM classification result, where the × point represents the normal state and the small circle point represents the abnormal state. The large circle point represents the support vector, i.e. the key sample near the decision boundary of the SVM model. These samples are the "key points" used to determine the decision boundary in the training process of the SVM classifier. Figure 4The middle X-axis and Y-axis show the first two features (Feature 1 and Feature 2) of the test dataset, and the scatter plot of the two features shows the distribution of data points in the feature space. The SVM finds a hyperplane (decision boundary) in the feature space by support vectors to separate the two classes of samples (normal and abnormal) as much as possible. Although the decision boundary is not directly plotted in the figure, the boundary of the classification can be inferred by observing the position of the support vector. If the support vector is located on the boundary of the two classes of data, it means that the model determines the classification boundary through these support vectors. The classification boundary will try to separate the normal samples from the abnormal samples as much as possible. The support vector is the most critical sample during the training of the SVM model, and it determines the final position of the classification boundary.

[0079] These points play a crucial role in the training and prediction of the model. Even samples far from the boundary do not affect the position of the classification boundary, only samples close to the boundary are support vectors. Most of the × points and small circles can be clearly separated, indicating that the SVM classification effect is good and can correctly distinguish between normal and abnormal states. If the × points and small circles overlap, the classification effect is poor, and the model may not be able to effectively distinguish between the two states.

[0080] In summary, the photovoltaic power generation system operation state evaluation method of the application obtains evaluation indexes of the photovoltaic power generation system to be evaluated, uses a subjective weighting method to subjectively weight the evaluation indexes to obtain subjective weights, uses multiple objective weighting methods to objectively weight the evaluation indexes to obtain objective weights, combines the subjective weights and the objective weights to obtain combined weights, inputs the evaluation indexes and the combined weights into a trained support vector machine model to perform operation state evaluation, and obtains an operation state evaluation result. The subjective and objective combined weighting method effectively fuses subjective experience and objective data-driven weight information, accurately quantifies the relative importance of each operation parameter, and the support vector machine model has strong nonlinear classification and generalization capabilities, which is particularly advantageous in processing small samples and high-dimensional data. It can accurately distinguish the operation state according to multiple parameter characteristics, so the combination of the subjective and objective combined weighting and the support vector machine model can efficiently and accurately evaluate the operation state of the photovoltaic power generation system. In addition, the radial basis kernel function is selected, the support vector machine model is trained using the training data set according to the radial basis kernel function, and the sequential minimal optimization algorithm is used to solve the model optimization problem, which can finely tune the model parameters and effectively improve the adaptability and reliability of the support vector machine model, providing a reliable basis for photovoltaic power generation system operation and maintenance, fault warning, and promoting the intelligent development of distributed photovoltaic power generation industry.

[0081] According to a further aspect of the application, Figure 2is a schematic diagram showing a photovoltaic power generation system operation state evaluation system according to an embodiment of the present application. The system comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements each step of the photovoltaic power generation system operation state evaluation method as described above when executing the computer program.

[0082] The above description is only an embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent transformation or direct or indirect application in the related technical field using the content of the present application specification and drawings is also included in the patent protection scope of the present application.

Claims

1. A method for evaluating the operating status of a photovoltaic power generation system, characterized in that, Including the following steps: Obtain the evaluation indicators for the photovoltaic power generation system to be evaluated; The evaluation indicators were subjectively weighted using a subjective weighting method to obtain subjective weights; The evaluation indicators are objectively weighted using multiple objective weighting methods to obtain objective weights; The subjective weights and the objective weights are combined and weighted to obtain the combined weights; The evaluation index and the combined weights are used to form a feature vector, which is then input into the trained support vector machine model to evaluate the operating status, thereby obtaining the operating status evaluation result of the photovoltaic power generation system to be evaluated.

2. The method for evaluating the operating status of a photovoltaic power generation system according to claim 1, characterized in that, Before obtaining the evaluation metrics for the photovoltaic power generation system to be evaluated, the following steps are also included: Define the radial basis kernel function for the support vector machine model; Obtain the training dataset; The support vector machine model is trained using the training dataset based on the radial basis function. The model optimization problem is solved using the sequence minimum optimization algorithm to obtain the parameters of the trained support vector machine model, thereby completing the training of the support vector machine model.

3. The method for evaluating the operating status of a photovoltaic power generation system according to claim 2, characterized in that, Define the radial basis kernel function of the support vector machine model as follows: ； In the formula, K ( x i , x j ) represents the radial basis function kernel. x i Representing the eigenvector i , x j Representing the eigenvector j , Indicates kernel parameters.

4. The method for evaluating the operating status of a photovoltaic power generation system according to claim 1, characterized in that, The evaluation indicators were subjectively weighted using a subjective weighting method, resulting in subjective weights including: The comparison matrix of the evaluation indicators was constructed using the three-scale analytic hierarchy process. Calculate the judgment matrix based on the comparison matrix; The subjective weights of the evaluation indicators are calculated based on the judgment matrix.

5. The method for evaluating the operating status of a photovoltaic power generation system according to claim 4, characterized in that, The comparison matrix of the evaluation indicators is constructed using the three-scale analytic hierarchy process, as follows: ； ； In the formula, A Represents a comparison matrix. n Indicates the number of evaluation indicators. a ij Indicators i With indicators j Compare matrix elements; The judgment matrix is ​​calculated based on the comparison matrix, specifically as follows: ； ； In the formula, D Represents the judgment matrix. d ij Indicators i With indicators j Determine the elements of the matrix. a ik Indicators i With indicators k Comparison of matrix elements, a kj Indicators k With indicators j Compare matrix elements; The subjective weights of the evaluation indicators are calculated based on the judgment matrix, specifically as follows: ； In the formula, a j Indicators j Subjective weighting.

6. The method for evaluating the operating status of a photovoltaic power generation system according to claim 1, characterized in that, The evaluation indicators are objectively weighted using multiple objective weighting methods, resulting in objective weights including: The evaluation indicators are objectively weighted using the mean square error weighting method to obtain the first objective weight; The evaluation indicators are objectively weighted using the grey relational analysis method to obtain the second objective weight. The objective weight is obtained by combining the first objective weight and the second objective weight.

7. The method for evaluating the operating status of a photovoltaic power generation system according to claim 6, characterized in that, The evaluation indicators are objectively weighted using the mean square error weighting method, resulting in the following first objective weights: Calculate the mean square error of the evaluation index; The mean squared error is normalized to obtain the first objective weight.

8. The method for evaluating the operating status of a photovoltaic power generation system according to claim 6, characterized in that, The evaluation indicators are objectively weighted using the grey relational analysis method, resulting in the following second objective weights: The maximum value of each column in the evaluation index is selected as the positive ideal solution, and the minimum value of each column in the evaluation index is selected as the negative ideal solution. The evaluation index and the positive ideal solution are calculated using Mahalanobis distance, as well as the evaluation index and the negative ideal solution. The degree of closeness between the evaluation index and the positive ideal solution is calculated based on the first distance and the second distance; The group grey relational degree of the evaluation index is calculated based on the proximity. The gray correlation degree of the group is normalized to obtain the second objective weight.

9. The method for evaluating the operating status of a photovoltaic power generation system according to claim 1, characterized in that, The subjective weights and objective weights are combined and weighted to obtain the combined weights, which include: A combined weighting model is constructed based on the subjective weights and the objective weights; The subjective and objective weight coefficients in the combined weighting model are solved using the Lagrange extreme value method. The subjective weight and the objective weight are combined and weighted according to the subjective weight coefficient and the objective weight coefficient to obtain the combined weight.

10. A photovoltaic power generation system operation status assessment system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the photovoltaic power generation system operation status assessment method according to any one of claims 1 to 9.

Images