Photovoltaic panel conversion power anomaly analysis method and system

Abstract

According to the photovoltaic panel conversion power anomaly analysis method and system provided by the invention, the similar candidate photovoltaic panel power conversion data is recalled from the plurality of reference photovoltaic panel power conversion data, on one hand, the recall efficiency is improved, and on the other hand, due to the fact that the stage type photovoltaic panel power conversion data has significant characteristics, the accuracy of the recall efficiency is improved. The characteristics of the power conversion data of the staged photovoltaic panel can be captured more accurately, and the recall accuracy is improved. Then, according to a second common weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed, selecting power abnormal conversion data from the candidate photovoltaic panel power conversion data; the power abnormal conversion data which is more similar to the photovoltaic panel power conversion data to be processed on the whole can be selected from the candidate photovoltaic panel power conversion data. Therefore, the precision and accuracy of abnormal power conversion analysis are improved.

Description

Technical Field

[0001] The present application relates to the technical field of abnormal data analysis, and more particularly, to a method and system for analyzing abnormal conversion power of photovoltaic panels. Background Art

[0002] The conversion power of a photovoltaic panel is understood as the photovoltaic panel utilizing the photovoltaic effect, that is, when photons strike a semiconductor material, electrons are released, thereby generating an electric current. In the actual operation process, the cleanliness and illumination conditions on the photovoltaic panel will affect the power conversion of the photovoltaic panel, and whether the components on the photovoltaic panel are complete will also affect the power conversion of the photovoltaic panel. Therefore, there is an urgent need for a method for analyzing abnormal conversion power of photovoltaic panels to improve the above technical problems. Summary of the Invention

[0003] To improve the technical problems existing in the related art, the present application provides a method and system for analyzing abnormal conversion power of photovoltaic panels.

[0004] In a first aspect, a method for analyzing abnormal conversion power of a photovoltaic panel is provided, including: Obtaining power conversion data of a photovoltaic panel to be processed; Performing key feature analysis processing on the power conversion data of the photovoltaic panel to be processed to obtain multiple stage-type power conversion data of the photovoltaic panel with significant features; For each stage-type power conversion data of the photovoltaic panel, obtaining multiple reference power conversion data of the photovoltaic panel corresponding to the stage-type power conversion data, and a first common weight between the stage-type power conversion data and each reference power conversion data; Combining the first common weight, determining one or more candidate power conversion data of the photovoltaic panel that match the stage-type power conversion data among the multiple reference power conversion data of the photovoltaic panel, and determining a second common weight of the candidate power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed; Combining the second common weight, determining power abnormal conversion data from the candidate power conversion data.

[0005] In the present application, the obtaining multiple reference power conversion data of the photovoltaic panel corresponding to the stage-type power conversion data includes: Performing feature extraction on the stage-type power conversion data of the photovoltaic panel to obtain power conversion data features of the photovoltaic panel; Obtaining a reference feature vector that matches the power conversion data features of the photovoltaic panel from a plurality of preset original feature vectors; the plurality of original feature vectors are extracted from a plurality of preset original power conversion data of the photovoltaic panel; Perform traceability processing on the reference feature vector for photovoltaic panel power conversion data to obtain the original photovoltaic panel power conversion data corresponding to the reference feature vector; Use the original photovoltaic panel power conversion data corresponding to the reference feature vector as the reference photovoltaic panel power conversion data.

[0006] In this application, obtaining the original feature vector that matches the characteristics of the photovoltaic panel power conversion data from a plurality of preset original feature vectors as the reference feature vector includes: Obtain the candidate common weights between the characteristics of the photovoltaic panel power conversion data and the plurality of original feature vectors respectively; Sort the plurality of original feature vectors in the order set by the candidate common weights to obtain the sorted original feature vectors; Use the first X original feature vectors in the sorted original feature vectors as the original feature vectors that match the characteristics of the photovoltaic panel power conversion data, where X is a positive integer; Use the original feature vector that matches the characteristics of the photovoltaic panel power conversion data as the reference feature vector.

[0007] In this application, obtaining the candidate common weights between the characteristics of the photovoltaic panel power conversion data and the plurality of original feature vectors respectively includes: For each original feature vector, perform quantization processing on the original feature vector to obtain the stage-type AI feature space corresponding to the original feature vector. The stage-type AI feature space includes a plurality of stage-type vectors with the same dimension, and the dimension of the stage-type vector is smaller than the dimension of the original feature vector; Perform discrimination processing on the stage-type AI feature space through a preset classification thread to obtain the discrimination center point of the stage-type AI feature space; Decode the stage-type AI feature space through the discrimination center point to obtain the type decoding corresponding to the stage-type AI feature space; Determine the candidate common weight between the characteristics of the photovoltaic panel power conversion data and the original feature vector through the type decoding and the characteristics of the photovoltaic panel power conversion data.

[0008] In this application, obtaining the first common weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data respectively includes: For the reference photovoltaic panel power conversion data, identify the number of vectors of the reference feature vector corresponding to the reference photovoltaic panel power conversion data; If the number of the vectors is one, calculate the commonality weight between the reference characteristic vector corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data characteristic corresponding to the staged photovoltaic panel power conversion data, to obtain a first commonality weight between the staged photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data; If the number of the vectors is multiple, respectively calculate the commonality weight between each of the multiple reference characteristic vectors corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data characteristic corresponding to the staged photovoltaic panel power conversion data, to obtain multiple original commonality weights, and fuse the multiple original commonality weights, to obtain a first commonality weight between the staged photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

[0009] In the present application, combining the first commonality weight to determine one or more candidate photovoltaic panel power conversion data that match the staged photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data includes: Sort the multiple reference photovoltaic panel power conversion data according to the order set by the first commonality weight, to obtain the sorted reference photovoltaic panel power conversion data; Use the first z reference photovoltaic panel power conversion data in the sorted reference photovoltaic panel power conversion data as the candidate photovoltaic panel power conversion data, where z is a positive integer.

[0010] In the present application, combining the first commonality weight to determine a second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the to-be-processed photovoltaic panel power conversion data includes: For each candidate photovoltaic panel power conversion data, fuse the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data, to obtain a second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the to-be-processed photovoltaic panel power conversion data.

[0011] In the present application, fusing the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data, to obtain a second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the to-be-processed photovoltaic panel power conversion data includes: Obtain the fusion weight of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data and the number of multiple staged photovoltaic panel power conversion data; Perform a weighting process on the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data through the fusion weight, to obtain a fused commonality weight; Calculate the comparison result between the fused common weight and the number of the multiple stage - type photovoltaic panel power conversion data to obtain the second common weight.

[0012] In this application, the key feature analysis process for the to - be - processed photovoltaic panel power conversion data to obtain multiple stage - type photovoltaic panel power conversion data with significant features includes: Identify the factor characters in the to - be - processed photovoltaic panel power conversion data, where the factor characters are segments with changing features in the to - be - processed photovoltaic panel power conversion data; Perform key feature analysis on the to - be - processed photovoltaic panel power conversion data through the factor characters to obtain multiple stage - type photovoltaic panel power conversion data with significant features.

[0013] In this application, the identification of the factor characters in the to - be - processed photovoltaic panel power conversion data includes: Extract features from the to - be - processed photovoltaic panel power conversion data to obtain a feature ranking, where each feature in the feature ranking corresponds to a segment in the to - be - processed photovoltaic panel power conversion data; Divide the feature ranking into a first local feature ranking and a second local feature ranking; Perform matrix calculations on the first local feature ranking, the second local feature ranking, and the feature ranking respectively to obtain a first matrix corresponding to the first local feature ranking, a second matrix corresponding to the second local feature ranking, and a third matrix corresponding to the feature ranking; Determine the maximum weight value according to the first matrix, the second matrix, and the third matrix; if the maximum weight value meets the preset condition, then use the segment corresponding to the last feature in the first local feature ranking as the factor character.

[0014] In this application, the determination of the power abnormal conversion data from the candidate photovoltaic panel power conversion data by combining the second common weight includes: Use the candidate photovoltaic panel power conversion data with the largest second common weight among all candidate photovoltaic panel power conversion data matched with the multiple reference photovoltaic panel power conversion data as the power abnormal conversion data.

[0015] In a second aspect, a photovoltaic panel conversion power abnormal analysis system is provided, including a processor and a memory that communicate with each other. The processor is configured to read and execute a computer program from the memory to implement the above - mentioned method.

[0016] A method and system for analyzing abnormal conversion power of a photovoltaic panel provided by an embodiment of the present application. After obtaining the power conversion data of the photovoltaic panel to be processed, the embodiment of the present application can split the power conversion data of the photovoltaic panel to be processed into multiple stage-type photovoltaic panel power conversion data with significant features, and then for each stage-type photovoltaic panel power conversion data, obtain a plurality of reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data, and the first common weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data; then, according to the first common weight, determine one or more candidate photovoltaic panel power conversion data that match the stage-type photovoltaic panel power conversion data among the plurality of reference photovoltaic panel power conversion data, and determine the second common weight of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed; finally, according to the second common weight, determine the power abnormal conversion data from the candidate photovoltaic panel power conversion data. In the embodiment of the present application, for each stage-type photovoltaic panel power conversion data of the power conversion data of the photovoltaic panel to be processed, some similar reference photovoltaic panel power conversion data can be initially recalled according to the stage-type photovoltaic panel power conversion data, and then according to the first common weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data, similar candidate photovoltaic panel power conversion data can be recalled from the plurality of reference photovoltaic panel power conversion data. On the one hand, the recall efficiency is improved, and on the other hand, since the stage-type photovoltaic panel power conversion data has significant features, the features of the stage-type photovoltaic panel power conversion data can be captured more accurately, and the recall accuracy is improved. Then, according to the second common weight of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed, the power abnormal conversion data is selected from the candidate photovoltaic panel power conversion data, and the power abnormal conversion data that is more similar to the power conversion data of the photovoltaic panel to be processed as a whole can be selected from the candidate photovoltaic panel power conversion data. Thus, the accuracy and precision of the power abnormal conversion analysis are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of a method for analyzing abnormal conversion power of a photovoltaic panel provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] To better understand the above technical solution, the technical solution of the present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0020] Please refer to Figure 1 , which shows a method for analyzing abnormal conversion power of a photovoltaic panel. The method may include the technical solutions described in the following steps 101-105.

[0021] 101. Obtain the power conversion data of the photovoltaic panel to be processed.

[0022] For example, the power conversion efficiency of a photovoltaic panel is generally between 15% and 25%. The specific data is as follows: Monocrystalline silicon solar cells: The conversion efficiency is usually between 20% and 23%. Polycrystalline silicon solar cells: The conversion efficiency is usually between 15% and 18%. Thin-film solar cells: The conversion efficiency is generally between 10% and 12%, but some advanced technologies can reach an efficiency of more than 15%. These data may change with technological progress. The power conversion data of the photovoltaic panel can be obtained through detection equipment.

[0023] 102. Perform key feature analysis and processing on the power conversion data of the photovoltaic panel to be processed to obtain multiple stage-type power conversion data of the photovoltaic panel with significant features.

[0024] Among them, the key feature analysis and processing refers to decomposing, extracting, etc. the parts with different features in the power conversion data of the photovoltaic panel.

[0025] Among them, the fact that the stage-type power conversion data of the photovoltaic panel has significant features indicates that the segment of the stage-type power conversion data of the photovoltaic panel is relatively independent in terms of features, that is, each stage-type power conversion data of the photovoltaic panel contains specific feature information or content by itself, and does not need to rely on the entire power conversion data of the photovoltaic panel to be processed or other stage-type power conversion data of the photovoltaic panel to understand its meaning.

[0026] Regarding a possible embodiment, in the case where the power conversion data of the photovoltaic panel to be processed does not have features, the key feature analysis and processing can also be performed on the power conversion data of the photovoltaic panel to be processed through the features of the power conversion data of the photovoltaic panel to obtain multiple stage-type power conversion data of the photovoltaic panel with significant features, that is, different stage-type power conversion data in the multiple stage-type power conversion data after splitting can have different features of the power conversion data of the photovoltaic panel.

[0027] For a possible embodiment, in step 102, the specific implementation of the step "perform key feature analysis on the photovoltaic panel power conversion data to be processed to obtain multiple stage-type photovoltaic panel power conversion data with significant features" may include: z1. Identify the factor characters in the photovoltaic panel power conversion data to be processed, where the factor characters are the segments where the features change in the photovoltaic panel power conversion data to be processed.

[0028] For a possible embodiment, the factor characters in the photovoltaic panel power conversion data to be recognized can be identified by a pre-trained factor character prediction thread. The factor character prediction thread can output corresponding factor characters according to the input photovoltaic panel power conversion data to be processed. Exemplarily, when training the factor character prediction thread, a large number of photovoltaic panel power conversion data examples can be prepared first, and then the factor characters in the photovoltaic panel power conversion data examples are marked. Then, the photovoltaic panel power conversion data examples marked with factor characters are input into the original thread for thread training, with the goal of enabling the original thread to accurately identify the factor characters in the photovoltaic panel power conversion data to obtain the factor character prediction thread.

[0029] In an alternative embodiment, in step z1, the specific implementation of the step "identify the factor characters in the photovoltaic panel power conversion data to be processed" may include: z11z. Extract features from the photovoltaic panel power conversion data to be processed to obtain a feature ranking, where each feature in the feature ranking corresponds to a segment in the photovoltaic panel power conversion data to be processed.

[0030] z12z. Divide the feature ranking into a first local feature ranking and a second local feature ranking.

[0031] z13z. Perform matrix calculations on the first local feature ranking, the second local feature ranking, and the feature ranking respectively to obtain a first matrix corresponding to the first local feature ranking, a second matrix corresponding to the second local feature ranking, and a third matrix corresponding to the feature ranking.

[0032] z14z. Determine the maximum weight value according to the first matrix, the second matrix, and the third matrix; if the maximum weight value meets the preset condition, the segment corresponding to the last feature in the first local feature ranking is used as the factor character.

[0033] In an alternative embodiment, in step z1, the specific implementation of the step "identify the factor characters in the photovoltaic panel power conversion data to be processed" may include: z11b. Extract conversion anomaly features from the photovoltaic panel power conversion data to be processed, obtaining multiple conversion anomaly features. Each conversion anomaly feature among the multiple conversion anomaly features corresponds to a segment in the photovoltaic panel power conversion data to be processed, and the multiple conversion anomaly features are arranged in the order of the segments from the earliest to the latest.

[0034] Among them, the quantity and dimension of the conversion anomaly features can be custom-set according to actual requirements and are not limited herein.

[0035] z12b. Determine one or more target conversion anomaly features from the multiple conversion anomaly features. The target conversion anomaly features are used to divide the multiple conversion anomaly features into multiple conversion anomaly feature sets, and the conversion anomaly features in each conversion anomaly feature set satisfy Gaussian distribution.

[0036] Regarding a possible embodiment, taking the conversion anomaly feature sorting composed of multiple conversion anomaly features as an example, the specific implementation of determining one or more target conversion anomaly features from the multiple conversion anomaly features may include: Selecting one conversion anomaly feature from the conversion anomaly feature sorting as a reference conversion anomaly feature.

[0037] Then, take the reference conversion anomaly feature and the conversion anomaly features in the conversion anomaly feature sorting that are before the reference conversion anomaly feature as the to-be-tested conversion anomaly feature set, and determine whether the to-be-tested conversion anomaly feature set satisfies Gaussian distribution.

[0038] If it is satisfied, the reference conversion anomaly feature can be used as the target conversion anomaly feature, and the reference conversion anomaly feature and the conversion anomaly features in the conversion anomaly feature sorting that are before the reference conversion anomaly feature are removed from the conversion anomaly feature sorting, obtaining a new conversion anomaly feature sorting, and return to execute the above step of "selecting one conversion anomaly feature from the conversion anomaly feature sorting as a reference conversion anomaly feature" based on the new conversion anomaly feature sorting until there is no to-be-tested conversion anomaly feature set that satisfies Gaussian distribution in the new conversion anomaly feature sorting.

[0039] If it is not satisfied, take the adjacent conversion anomaly feature after the reference conversion anomaly feature in the conversion anomaly feature sorting as the new reference conversion anomaly feature, and return to execute the above step of "taking the reference conversion anomaly feature and the conversion anomaly features in the conversion anomaly feature sorting that are before the reference conversion anomaly feature as the to-be-tested conversion anomaly feature set, and determining whether the to-be-tested conversion anomaly feature set satisfies Gaussian distribution" based on the new reference conversion anomaly feature until the to-be-tested conversion anomaly feature set satisfies Gaussian distribution.

[0040] In this embodiment, by selecting conversion abnormality features one by one from multiple conversion abnormality features as reference conversion abnormality features, and constructing a ranking of conversion abnormality features to be tested based on the reference conversion abnormality features, and then identifying whether the accurate ranking of the conversion abnormality features to be tested satisfies the Gaussian arrangement to determine whether the reference conversion abnormality feature is the target conversion abnormality feature, the target conversion abnormality feature can be accurately determined from the multiple conversion abnormality features, avoiding the omission of the target conversion abnormality feature.

[0041] In an alternative embodiment, in step z12b, the specific implementation of the step of "determining one or more target conversion abnormality features from multiple conversion abnormality features" may include: z121b. Determine a candidate conversion exception feature from multiple conversion exception features.

[0042] In relation to a possible implementation example, a conversion abnormality feature may be randomly selected from a plurality of conversion abnormality features as a candidate conversion abnormality feature, wherein when selecting the candidate conversion abnormality feature, the first conversion abnormality feature and the last conversion abnormality feature in the plurality of conversion abnormality features may be ignored. Exemplarily, taking a conversion abnormality feature ranking composed of a plurality of conversion abnormality features as an example, when selecting the candidate conversion abnormality feature, the conversion abnormality feature close to the middle in the conversion abnormality feature ranking may be preferentially selected as the candidate conversion abnormality feature.

[0043] z122b. Differentiate multiple conversion abnormality features into a first conversion abnormality feature set and a second conversion abnormality feature set according to the candidate conversion abnormality features; the first conversion abnormality feature set includes the candidate conversion abnormality features and conversion abnormality features whose fragments are before the candidate conversion abnormality features; the second conversion abnormality feature set includes conversion abnormality features whose fragments are after the candidate conversion abnormality features.

[0044] z123b. Determine evaluation values corresponding to candidate conversion abnormality features according to the first conversion abnormality feature set, the second conversion abnormality feature set, and the plurality of conversion abnormality features; the evaluation values represent the possibility that both the first conversion abnormality feature set and the second conversion abnormality feature set satisfy the Gaussian arrangement.

[0045] Specifically, in step z123b, according to the first conversion abnormality feature set, the second conversion abnormality feature set, and the plurality of conversion abnormality features, a specific implementation method for determining the evaluation value corresponding to the candidate conversion abnormality feature may include: z1231b. Determine the maximum weight value corresponding to the candidate conversion abnormality feature according to the first conversion abnormality feature set, the second conversion abnormality feature set and the multiple conversion abnormality features, wherein the maximum weight value represents the original possibility that the first conversion abnormality feature set and the second conversion abnormality feature set both satisfy the Gaussian arrangement.

[0046] For a possible embodiment, the first transformed abnormal feature set, the second transformed abnormal feature set, and a plurality of transformed abnormal features can be calculated respectively through the calculation formula of the matrix to obtain the first matrix corresponding to the first transformed abnormal feature set, the second matrix corresponding to the second transformed abnormal feature set, and the third matrix corresponding to the plurality of transformed abnormal features.

[0047] Then, perform det calculations on the first matrix, the second matrix, and the third matrix respectively to obtain the first det corresponding to the first matrix, the second det corresponding to the second matrix, and the third det corresponding to the third matrix.

[0048] Then, take the product of the logarithm of the first det and the number of transformed abnormal features in the first transformed abnormal feature set as the first calculated value, take the product of the logarithm of the second det and the number of transformed abnormal features in the second transformed abnormal feature set as the second calculated value, and take the product of the logarithm of the third det and the number of transformed abnormal features in the plurality of transformed abnormal features as the third calculated value.

[0049] Finally, subtract the first calculated value from the third calculated value and then subtract the second calculated value to obtain the maximum weight value corresponding to the candidate transformed abnormal feature.

[0050] z1232b. Obtain the number of features of a plurality of transformed abnormal features and the feature dimension of the transformed abnormal features, and determine an outlier according to the number of features and the feature dimension. The outlier is used to debug the original possibility.

[0051] z1233b. Determine an evaluation value according to the maximum weight value and the outlier.

[0052] For a possible embodiment, the weight corresponding to the outlier can be obtained, then calculate the product of the weight and the outlier to obtain a fourth calculated value, and subtract the fourth calculated value from the maximum weight value to obtain the evaluation value.

[0053] z124b. If the evaluation value is the maximum value in the corresponding value range of the evaluation value, determine the candidate transformed abnormal feature as the target transformed abnormal feature.

[0054] Wherein, when the evaluation value is the maximum value in the corresponding value range of the evaluation value, it indicates that the first transformed abnormal feature set and the second transformed abnormal feature set most likely satisfy the Gaussian distribution, that is, the segment corresponding to the candidate transformed abnormal feature that divides the first transformed abnormal feature set and the second transformed abnormal feature set is most likely a factor character.

[0055] In this embodiment, by converting the process of finding factor characters into the process of selecting the following two hypothetical threads, and then using the Bayesian information criterion for thread selection, the target conversion anomaly feature corresponding to the factor character can be quickly found among multiple conversion anomaly features, thereby improving the determination efficiency of the target conversion anomaly feature.

[0056] z13b. Use the segment corresponding to the target conversion anomaly feature as the factor character.

[0057] z2. Based on the factor character, perform key feature analysis and processing on the photovoltaic panel power conversion data to be processed, and obtain multiple stage-type photovoltaic panel power conversion data with significant features.

[0058] Among them, the key feature is understood as a feature that has a great impact on the photovoltaic panel power conversion data.

[0059] 103. For each stage-type photovoltaic panel power conversion data, obtain multiple reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data, and the first commonality weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data.

[0060] Among them, the reference photovoltaic panel power conversion data can be the central point photovoltaic panel power conversion data used for comparison with the photovoltaic panel power conversion data to be processed. Exemplarily, compared with the photovoltaic panel power conversion data to be processed, the central point photovoltaic panel power conversion data can be known, complete, and photovoltaic panel power conversion data without noise. Optionally, the reference photovoltaic panel power conversion data can be pre-stored in the above photovoltaic panel power conversion database.

[0061] Among them, each stage-type photovoltaic panel power conversion data can establish a mapping relationship with multiple reference photovoltaic panel power conversion data in advance, and the reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data can be photovoltaic panel power conversion data with similar features to the stage-type photovoltaic panel power conversion data.

[0062] Regarding a possible embodiment, in step 103, the specific implementation of obtaining multiple reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data may include: S1. Extract features from the stage-type photovoltaic panel power conversion data to obtain photovoltaic panel power conversion data features.

[0063] Regarding a possible embodiment, the stage-type photovoltaic panel power conversion data can be subjected to feature extraction by a preset photovoltaic panel power conversion data feature extraction unit. Optionally, the extracted photovoltaic panel power conversion data features can be matched with the above reference feature vectors in terms of feature dimensions to facilitate the calculation of the commonality weight between the two.

[0064] S2. Obtain a reference feature vector by acquiring the original feature vector that matches the photovoltaic panel power conversion data feature from a plurality of preset original feature vectors; the plurality of original feature vectors are extracted from a plurality of preset original photovoltaic panel power conversion data.

[0065] For a possible embodiment, the specific implementation of step S2 may include: S21. Obtain the candidate common weights between the photovoltaic panel power conversion data feature and each of the plurality of original feature vectors.

[0066] For a possible embodiment, the common weights between the photovoltaic panel power conversion data feature and each of the plurality of original feature vectors can be directly compared to obtain the candidate common weights.

[0067] In an alternative embodiment, the specific implementation of obtaining the candidate common weights between the photovoltaic panel power conversion data feature and each of the plurality of original feature vectors in S21 may include: S211. For each original feature vector, perform quantization processing on the original feature vector to obtain a stage-type AI feature space corresponding to the original feature vector. The stage-type AI feature space includes a plurality of stage-type vectors with the same dimension, and the dimension of the stage-type vector is smaller than the dimension of the original feature vector.

[0068] S212. Perform discrimination processing on the stage-type AI feature space through a preset classification thread to obtain the discrimination center point of the stage-type AI feature space.

[0069] S213. Decode the stage-type AI feature space based on the discrimination center point to obtain the type decoding corresponding to the stage-type AI feature space.

[0070] S214. Determine the candidate common weights between the photovoltaic panel power conversion data feature and the original feature vector based on the type decoding and the photovoltaic panel power conversion data feature.

[0071] Optionally, the discrimination center point in the stage-type AI feature space can be preferentially selected through the type decoding to calculate the common weights with the photovoltaic panel power conversion data feature, and then, compare all the vectors in the stage-type AI feature space to obtain the vectors similar to the photovoltaic panel power conversion data feature. Among them, the common weights can be calculated by the way of vector inner product.

[0072] Optionally, the photovoltaic panel power conversion data feature can also be divided into stage-type vectors of photovoltaic panel power conversion data, and then calculate the sum of the differences between each stage-type vector of photovoltaic panel power conversion data and the corresponding quantized vector of each stage-type AI feature space to obtain the candidate common weights.

[0073] S22. Sort the multiple original feature vectors in the order of the set candidate common weights to obtain the sorted original feature vectors.

[0074] S23. Use the first X original feature vectors in the sorted original feature vectors as the original feature vectors that match the characteristics of the photovoltaic panel power conversion data, where X is a positive integer.

[0075] S24. Use the original feature vectors that match the characteristics of the photovoltaic panel power conversion data as the reference feature vectors.

[0076] S3. Perform traceability processing on the reference feature vectors for the photovoltaic panel power conversion data to obtain the original photovoltaic panel power conversion data corresponding to the reference feature vectors.

[0077] Among them, the preset multiple original feature vectors can be stored in a preset vector library. The preset vector library stores multiple original feature vectors and a vector mapping relationship table in advance. The vector mapping relationship table pre-records the original photovoltaic panel power conversion data corresponding to each original feature vector among the multiple original feature vectors. The original feature vectors are extracted from their corresponding original photovoltaic panel power conversion data. Therefore, after determining the original photovoltaic panel power conversion data, the original feature vectors corresponding to the original photovoltaic panel power conversion data can be found from the preset vector library according to the vector mapping relationship table. Among them, one original photovoltaic panel power conversion data can correspond to one or more original feature vectors.

[0078] S4. Use the original photovoltaic panel power conversion data corresponding to the reference feature vectors as the reference photovoltaic panel power conversion data.

[0079] Regarding a possible embodiment, in step 103, the specific implementation manner of obtaining the first common weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data may include: for the reference photovoltaic panel power conversion data, identify the number of vectors of the reference feature vectors corresponding to the reference photovoltaic panel power conversion data.

[0080] If the number of vectors is one, calculate the common weight between the reference feature vector corresponding to the reference photovoltaic panel power conversion data and the power conversion data feature of the photovoltaic panel corresponding to the stage-type photovoltaic panel power conversion data to obtain the first common weight between the stage-type photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

[0081] If the number of vectors is multiple, the commonality weights are calculated for each of the multiple reference feature vectors corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data features corresponding to the phased photovoltaic panel power conversion data to obtain multiple original commonality weights, and the multiple original commonality weights are fused to obtain the first commonality weight between the phased photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

[0082] Exemplarily, the specific way to obtain the first commonality weight can be as follows: B1. Obtain each reference feature vector corresponding to the reference photovoltaic panel power conversion data from the preset vector library.

[0083] The reference photovoltaic panel power conversion data may correspond to one or more reference feature vectors.

[0084] B2. Extract features from the phased photovoltaic panel power conversion data to obtain the photovoltaic panel power conversion data features.

[0085] B3. For each reference photovoltaic panel power conversion data, calculate the commonality weight between the reference feature vector corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data features to obtain the first commonality weight between the phased photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

[0086] When the number of reference feature vectors corresponding to the reference photovoltaic panel power conversion data is multiple, in step B3, the specific implementation of calculating the commonality weight between the reference feature vector corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data features to obtain the first commonality weight between the phased photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data may include: calculating the commonality weight between the photovoltaic panel power conversion data features and each of the multiple reference feature vectors corresponding to the reference photovoltaic panel power conversion data to obtain multiple original commonality weights.

[0087] Fuse the multiple original commonality weights to obtain the first commonality weight between the phased photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

[0088] 104. Determine one or more candidate photovoltaic panel power conversion data that match the phased photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data according to the first commonality weight, and determine the second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed.

[0089] For a possible embodiment, in step 104, the specific implementation manner of determining one or more candidate photovoltaic panel power conversion data that match the staged photovoltaic panel power conversion data from multiple reference photovoltaic panel power conversion data according to the first commonality weight may include: using the reference photovoltaic panel power conversion data with the first commonality weight greater than or equal to the commonality weight threshold among the multiple reference photovoltaic panel power conversion data as the candidate photovoltaic panel power conversion data.

[0090] In an alternative embodiment, in step 104, the specific implementation manner of determining one or more candidate photovoltaic panel power conversion data that match the staged photovoltaic panel power conversion data from multiple reference photovoltaic panel power conversion data according to the first commonality weight may include: sorting the multiple reference photovoltaic panel power conversion data according to the order set by the first commonality weight to obtain the sorted reference photovoltaic panel power conversion data.

[0091] Taking the first z reference photovoltaic panel power conversion data in the sorted reference photovoltaic panel power conversion data as the candidate photovoltaic panel power conversion data, where z is a positive integer.

[0092] For a possible embodiment, the reference photovoltaic panel power conversion data may also be directly used as the candidate photovoltaic panel power conversion data.

[0093] For a possible embodiment, in step 104, the specific implementation manner of determining the second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed according to the first commonality weight may include: for each candidate photovoltaic panel power conversion data, fusing the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data to obtain the second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed.

[0094] For a possible embodiment, the specific implementation manner of the step "fusing the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data to obtain the second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed" may include: obtaining the fusion weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data and the number of multiple staged photovoltaic panel power conversion data; based on the fusion weights, performing a weighted process on the first commonality weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data to obtain the fused commonality weight; calculating the comparison result between the fused commonality weight and the number of multiple staged photovoltaic panel power conversion data to obtain the second commonality weight.

[0095] 105. Determine the power abnormal conversion data from the candidate PV panel power conversion data according to the second commonality weight.

[0096] For a possible embodiment, in step 105, the specific implementation of determining the power abnormal conversion data from the candidate PV panel power conversion data according to the second commonality weight may include: taking the candidate PV panel power conversion data with the largest second commonality weight among all the candidate PV panel power conversion data matched with multiple reference PV panel power conversion data as the power abnormal conversion data.

[0097] For a possible embodiment, after determining the power abnormal conversion data, the power abnormal conversion data may also be output. Optionally, in addition to outputting the power abnormal conversion data, the PV panel power conversion data information corresponding to the power abnormal conversion data may also be output.

[0098] It can be seen that in this embodiment, after obtaining the to-be-processed PV panel power conversion data, the to-be-processed PV panel power conversion data is split into multiple stage-type PV panel power conversion data with significant features. Then, for each stage-type PV panel power conversion data, multiple reference PV panel power conversion data corresponding to the stage-type PV panel power conversion data and the first commonality weight between the stage-type PV panel power conversion data and each reference PV panel power conversion data are obtained. Then, according to the first commonality weight, one or more candidate PV panel power conversion data matched with the stage-type PV panel power conversion data among the multiple reference PV panel power conversion data are determined, and the second commonality weight of the candidate PV panel power conversion data corresponding to the to-be-processed PV panel power conversion data is determined. Finally, according to the second commonality weight, the power abnormal conversion data is determined from the candidate PV panel power conversion data. Since for each stage-type PV panel power conversion data of the to-be-processed PV panel power conversion data, similar candidate PV panel power conversion data are recalled from the multiple reference PV panel power conversion data according to the first commonality weight between the stage-type PV panel power conversion data and each reference PV panel power conversion data, on the one hand, the recall efficiency is improved, and on the other hand, because the stage-type PV panel power conversion data has significant features, the features of the stage-type PV panel power conversion data can be captured more accurately, and the recall accuracy is improved. Then, according to the second commonality weight of the candidate PV panel power conversion data corresponding to the to-be-processed PV panel power conversion data, the power abnormal conversion data is selected from the candidate PV panel power conversion data, and the power abnormal conversion data that is more similar to the to-be-processed PV panel power conversion data as a whole can be selected from the candidate PV panel power conversion data. Thus, the accuracy, efficiency of the power abnormal conversion analysis are improved.

[0099] On the basis described above, a photovoltaic panel conversion power anomaly analysis device is provided. The device includes: A data acquisition module, configured to acquire the power conversion data of the photovoltaic panel to be processed; A data obtaining module, configured to perform key feature analysis processing on the power conversion data of the photovoltaic panel to be processed, and obtain multiple stage-type photovoltaic panel power conversion data with significant features; A first weight acquisition module, configured to, for each stage-type photovoltaic panel power conversion data, acquire multiple reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data, and the first common weights between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data; A second weight acquisition module, configured to combine the first common weights to determine one or more candidate photovoltaic panel power conversion data that match the stage-type photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data, and determine the second common weights of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed; A problem data determination module, configured to combine the second common weights to determine power anomaly conversion data from the candidate photovoltaic panel power conversion data.

[0100] On the basis described above, a photovoltaic panel conversion power anomaly analysis system is shown, including a processor and a memory that communicate with each other. The processor is configured to read and execute a computer program from the memory to implement the above method.

[0101] On the basis described above, a computer-readable storage medium is further provided, and a computer program stored thereon implements the above method when running.

[0102] In summary, based on the above solution, in the embodiment of the present application, after acquiring the power conversion data of the photovoltaic panel to be processed, the power conversion data of the photovoltaic panel to be processed can be split into multiple stage-type photovoltaic panel power conversion data with significant features, and then, for each stage-type photovoltaic panel power conversion data, multiple reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data, and the first common weights between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data are acquired; then, according to the first common weights, one or more candidate photovoltaic panel power conversion data that match the stage-type photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data are determined, and the second common weights of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed are determined; finally, according to the second common weights, power anomaly conversion data is determined from the candidate photovoltaic panel power conversion data.

[0103] In the embodiments of the present application, for each stage-type photovoltaic panel power conversion data to be processed, some similar reference photovoltaic panel power conversion data can be initially recalled according to the stage-type photovoltaic panel power conversion data. Then, according to the first commonality weight between the stage-type photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data, similar candidate photovoltaic panel power conversion data can be recalled from multiple reference photovoltaic panel power conversion data. On the one hand, the recall efficiency is improved. On the other hand, since the stage-type photovoltaic panel power conversion data has significant features, the features of the stage-type photovoltaic panel power conversion data can be captured more accurately, improving the recall accuracy. Then, according to the second commonality weight of the candidate photovoltaic panel power conversion data corresponding to the photovoltaic panel power conversion data to be processed, power abnormal conversion data can be selected from the candidate photovoltaic panel power conversion data, and power abnormal conversion data that is more similar to the photovoltaic panel power conversion data to be processed as a whole can be selected from the candidate photovoltaic panel power conversion data. Thus, the accuracy and precision of power abnormal conversion analysis are improved.

[0104] It should be understood that the above-described systems and their modules can be implemented in various ways. For example, in some embodiments, the systems and their modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art can understand that the above methods and systems can be implemented using computer-executable instructions and / or included in processor control code, such as provided on a carrier medium such as a disk, CD, or DVD-ROX, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The systems and their modules of the present application can be implemented not only by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but also by software implemented by various types of processors, or by a combination of the above hardware circuits and software (e.g., firmware).

[0105] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced can be any one or several combinations of the above, or any other beneficial effects that may be obtained.

Claims

1. A method for analyzing abnormal conversion power of a photovoltaic panel, characterized in that, Including: Obtain the power conversion data of the photovoltaic panel to be processed; Perform key feature analysis processing on the power conversion data of the photovoltaic panel to be processed, and obtain multiple staged photovoltaic panel power conversion data with significant features; For each staged photovoltaic panel power conversion data, obtain multiple reference photovoltaic panel power conversion data corresponding to the staged photovoltaic panel power conversion data, and the first common weight between the staged photovoltaic panel power conversion data and each reference photovoltaic panel power conversion data; wherein, the first common weight is used to represent the first similarity, and each reference photovoltaic panel power conversion data is understood as example data, representing the similarity between the staged photovoltaic panel power conversion data and the example data; Combined with the first common weight, determine one or more candidate photovoltaic panel power conversion data that match the staged photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data, and determine the second common weight of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed; wherein, the second common weight is used to represent the second similarity, and is used to determine the ecological problems in the power conversion data of the photovoltaic panel to be processed; Combined with the second common weight, determine the power abnormal conversion data from the candidate photovoltaic panel power conversion data.

2. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 1, wherein The combining the first common weight to determine one or more candidate photovoltaic panel power conversion data that match the staged photovoltaic panel power conversion data among the multiple reference photovoltaic panel power conversion data includes: Sort the multiple reference photovoltaic panel power conversion data according to the order set by the first common weight, and obtain the sorted reference photovoltaic panel power conversion data; Take the first z reference photovoltaic panel power conversion data in the sorted reference photovoltaic panel power conversion data as the candidate photovoltaic panel power conversion data, where z is a positive integer; Wherein, combining the first common weight to determine the second common weight of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed includes: For each candidate photovoltaic panel power conversion data, fuse the first common weights of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data to obtain the second common weight of the candidate photovoltaic panel power conversion data corresponding to the power conversion data of the photovoltaic panel to be processed.

3. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 1, wherein Performing key feature analysis processing on the power conversion data of the photovoltaic panel to be processed to obtain multiple staged photovoltaic panel power conversion data with significant features includes: Identify the factor characters in the power conversion data of the photovoltaic panel to be processed, and the factor characters are the segments where the features in the power conversion data of the photovoltaic panel to be processed change; Based on the factor characters, perform key feature analysis processing on the power conversion data of the photovoltaic panel to be processed, and obtain multiple staged photovoltaic panel power conversion data with significant features.

4. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 1, wherein Identifying the factor characters in the power conversion data of the photovoltaic panel to be processed includes: Extract conversion anomaly features from the photovoltaic panel power conversion data to be processed, obtaining multiple conversion anomaly features. Each conversion anomaly feature among the multiple conversion anomaly features corresponds to a segment in the photovoltaic panel power conversion data to be processed, and the multiple conversion anomaly features are arranged in the order of the segments from the earliest to the latest; Determine one or more target conversion anomaly features from the multiple conversion anomaly features. The target conversion anomaly features are used to divide the multiple conversion anomaly features into multiple conversion anomaly feature sets, and the conversion anomaly features in each conversion anomaly feature set satisfy Gaussian distribution.

5. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 1, wherein The obtaining of the multiple reference photovoltaic panel power conversion data corresponding to the stage-type photovoltaic panel power conversion data includes: Extract features from the stage-type photovoltaic panel power conversion data to obtain photovoltaic panel power conversion data features; Obtain the original feature vector that matches the photovoltaic panel power conversion data features from a preset multiple of original feature vectors; the multiple original feature vectors are extracted from a preset multiple of original photovoltaic panel power conversion data; Perform traceability processing on the reference feature vector for photovoltaic panel power conversion data to obtain the original photovoltaic panel power conversion data corresponding to the reference feature vector; Use the original photovoltaic panel power conversion data corresponding to the reference feature vector as the reference photovoltaic panel power conversion data.

6. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 4, characterized in that, The obtaining of the original feature vector that matches the photovoltaic panel power conversion data features from a preset multiple of original feature vectors includes: Obtain the candidate commonality weights between the photovoltaic panel power conversion data features and the multiple original feature vectors respectively; Sort the multiple original feature vectors according to the order set by the candidate commonality weights to obtain the sorted original feature vectors; Use the first X original feature vectors in the sorted original feature vectors as the original feature vectors that match the photovoltaic panel power conversion data features, where X is a positive integer; Use the original feature vectors that match the photovoltaic panel power conversion data features as the reference feature vectors.

7. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 6, wherein The obtaining of the candidate commonality weights between the photovoltaic panel power conversion data features and the multiple original feature vectors respectively includes: For each original feature vector, perform quantization processing on the original feature vector to obtain the stage-type AI feature space corresponding to the original feature vector. The stage-type AI feature space includes multiple stage-type vectors with the same number of dimensions, and the number of dimensions of the stage-type vector is less than the number of dimensions of the original feature vector; Perform discrimination processing on the stage-type AI feature space through a preset classification thread to obtain the discrimination center point of the stage-type AI feature space; Decode the stage-type AI feature space through the discrimination center point to obtain the type decoding corresponding to the stage-type AI feature space; Determine the candidate commonality weights between the photovoltaic panel power conversion data features and the original feature vector through the type decoding and the photovoltaic panel power conversion data features; among them, the candidate commonality weights are used to represent the similarity of the power conversion description information in the initial data.

8. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 1, characterized in that, Obtaining the first common weight between the obtained power conversion data of the staged photovoltaic panel and each piece of reference photovoltaic panel power conversion data includes: For the reference photovoltaic panel power conversion data, identifying the number of vectors of the reference feature vector corresponding to the reference photovoltaic panel power conversion data; If the number of vectors is one, calculating the common weight between the reference feature vector corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data feature corresponding to the staged photovoltaic panel power conversion data to obtain the first common weight between the staged photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data; If the number of vectors is multiple, calculating the common weight between each of the multiple reference feature vectors corresponding to the reference photovoltaic panel power conversion data and the photovoltaic panel power conversion data feature corresponding to the staged photovoltaic panel power conversion data to obtain multiple original common weights, and fusing the multiple original common weights to obtain the first common weight between the staged photovoltaic panel power conversion data and the reference photovoltaic panel power conversion data.

9. The method for analyzing abnormal conversion power of a photovoltaic panel according to claim 2, wherein, Fusing the first common weight of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data to obtain the second common weight of the candidate photovoltaic panel power conversion data corresponding to the to-be-processed photovoltaic panel power conversion data includes: Obtaining the fusion weight of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data and the number of multiple staged photovoltaic panel power conversion data; Performing weighted processing on the first common weight of the candidate photovoltaic panel power conversion data corresponding to each staged photovoltaic panel power conversion data through the fusion weight to obtain a fused common weight; Calculating the comparison result between the fused common weight and the number of the multiple staged photovoltaic panel power conversion data to obtain the second common weight.

10. A photovoltaic panel conversion power anomaly analysis system, characterized in that, Including a processor and a memory that communicate with each other, the processor is configured to read and execute a computer program from the memory to implement the method according to any one of claims 1-9.

Images