Photovoltaic fuse life prediction method based on multi-parameter fusion

Abstract

The invention relates to the technical field of photovoltaic fuse life prediction, and discloses a photovoltaic fuse life prediction method based on multi-parameter fusion, and the method comprises the steps: determining the basic information of a photovoltaic fuse, obtaining related monitoring logs, obtaining photovoltaic environment factors in each related monitoring log, carrying out the classification, obtaining a plurality of monitoring log sets, and carrying out the collection of the monitoring logs; multi-source monitoring parameters of each related monitoring log are extracted and analyzed, and feature monitoring parameters of different feature photovoltaic environment factors and corresponding feature coefficients are determined according to an analysis result; setting a weight coefficient corresponding to the feature monitoring parameter according to the feature coefficient, and constructing a life prediction model; and determining a real-time photovoltaic environment change factor, generating a life prediction value based on the life prediction model, the real-time environment change factor and the real-time characteristic monitoring parameter, judging whether to generate an alarm signal, determining the influence weight of the multi-source monitoring parameter on the life value, and performing parameter fusion, thereby improving the life prediction precision.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic fuse life prediction, and in particular to a photovoltaic fuse life prediction method based on multi-parameter fusion. Background Art

[0002] Photovoltaic fuses are key protection devices in solar power generation systems, used to prevent line overload and short-circuit current from damaging photovoltaic modules and inverters. Photovoltaic fuse life prediction plays a key role in the design, operation and maintenance, and economic benefit optimization of photovoltaic power generation systems.

[0003] In the existing technology, a single sensor is usually used to collect monitoring parameters of photovoltaic fuses and perform life prediction. This lacks comprehensive monitoring data and does not consider the impact of multi-source monitoring parameters on life, resulting in low life prediction accuracy. Summary of the Invention

[0004] In order to solve the above technical problems, the present application provides a photovoltaic fuse life prediction method based on multi-parameter fusion. By determining several characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors, the characteristic monitoring parameters and corresponding weight coefficients under different photovoltaic environmental factors are obtained, and a life prediction model is constructed. The influence weights of multi-source monitoring parameters on the life value are determined and parameter fusion is performed to improve the accuracy of life prediction.

[0005] In some embodiments of the present application, a photovoltaic fuse life prediction method based on multi-parameter fusion is provided, including: Determine basic information of the photovoltaic fuse and obtain relevant monitoring logs, obtain and classify photovoltaic environmental factors in each relevant monitoring log, and obtain multiple monitoring log sets, wherein the monitoring log sets include several monitoring log subsets; Extract and analyze the multi-source monitoring parameters of each relevant monitoring log in several monitoring log subsets, and determine the characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors based on the analysis results; The weight coefficient of the corresponding characteristic monitoring parameter is set according to the characteristic coefficient, and the life prediction model is constructed according to several characteristic monitoring parameters of different photovoltaic environmental factors, the corresponding weight coefficient and several preset life values; Determine the real-time photovoltaic environment change factor, generate a life prediction value based on the life prediction model, the real-time environment change factor and the real-time characteristic monitoring parameters, and determine whether to generate an alarm signal.

[0006] In some embodiments of the present application, the photovoltaic environmental factors in each relevant monitoring log are obtained and classified to obtain multiple monitoring log sets, including: Obtain the historical total life value in each relevant monitoring log, and determine a number of monitoring stages corresponding to the relevant monitoring log according to the historical total life value and the preset monitoring interval; Pre-set a number of environmental monitoring and evaluation indicators; Obtain several historical environmental parameters for each monitoring phase in the same relevant monitoring log; Evaluate several historical environmental parameters based on several environmental monitoring and evaluation indicators, obtain reference evaluation values ​​of several environmental monitoring and evaluation indicators in each monitoring stage, and set them as photovoltaic environmental factors corresponding to relevant monitoring logs; The photovoltaic environmental factors of different relevant monitoring logs were analyzed for similarity, and similarity coefficients of different relevant monitoring logs were obtained; A monitoring log set is constructed based on related monitoring logs whose similarity coefficients are greater than a preset similarity coefficient threshold, and the same monitoring log set is further divided according to the corresponding total historical life values ​​to obtain several monitoring log subsets.

[0007] In some embodiments of the present application, obtaining similarity coefficients of different related monitoring logs includes: Randomly select a relevant monitoring log and set it as the target relevant monitoring log; Compare the photovoltaic environmental factors of each monitoring stage of the target-related monitoring log with the photovoltaic environmental factors of the corresponding monitoring stage of each related monitoring log, and obtain the sub-similarity coefficients of each monitoring stage of the target-related monitoring log and the corresponding monitoring stage of the corresponding related monitoring log; The calculation formula of the sub-similarity coefficient is: ; Among them, P1 is the sub-similarity coefficient, p0 is the similarity conversion coefficient, and n1 is the total number of environmental monitoring and evaluation indicators. is the reference evaluation value of the i1th environmental monitoring evaluation indicator of the target-related monitoring log in the current monitoring phase, is the reference evaluation value of the i1th environmental monitoring evaluation indicator in the corresponding monitoring log in the corresponding monitoring stage, and ai is the weight coefficient of the i-th environmental monitoring evaluation indicator; Generate a similarity coefficient between the target-related monitoring log and the corresponding related monitoring log according to several sub-similarity coefficients of all monitoring stages of the target-related monitoring log and the corresponding monitoring stages of the same related monitoring log; The calculation formula of the similarity coefficient is: ; Among them, P2 is the similarity coefficient, is the sub-similarity coefficient of the target related monitoring log and the corresponding related monitoring log in the i2th monitoring stage, is the sub-similarity coefficient threshold, is the weight coefficient of the i2th monitoring stage of the target-related monitoring log, and n2 is the total number of monitoring stages of the target-related monitoring log; Generate the similarity coefficient between the target related monitoring log and each related monitoring log in turn.

[0008] In some embodiments of the present application, extracting and analyzing multi-source monitoring parameters of each relevant monitoring log in several monitoring log subsets includes: Obtain the historical life span change value of each monitoring stage of each relevant monitoring log in the monitoring log subset, and set a number of extraction time nodes of the corresponding monitoring stage according to the historical life span change value and the corresponding preset time interval; Extract the multi-source monitoring parameters of the corresponding monitoring stage of each relevant monitoring log according to the extraction time node to obtain the multi-source monitoring parameter matrix W of each relevant monitoring log;

[0009] in, The historical monitoring parameters at the s1th extraction time node of the s2th parameter type of the relevant monitoring log, s1=1,…r, s2=1,…m; Analyze the multi-source monitoring parameter sequence W of each relevant monitoring log to determine a number of first undetermined characteristic monitoring parameters and corresponding first characteristic coefficients corresponding to the relevant monitoring log; Determine a second undetermined feature monitoring parameter and a corresponding second feature coefficient of a corresponding monitoring log subset according to a plurality of first undetermined feature monitoring parameters and corresponding first feature coefficients determined by different related monitoring logs of the same monitoring log subset; Constructing a plurality of monitoring log subset comparison combinations of the same monitoring log set and a plurality of comparison subcombinations of each monitoring log subset comparison combination; Analyze several comparison subcombinations to determine the third undetermined characteristic monitoring parameter and the corresponding third characteristic coefficient of each monitoring log subset comparison combination; Determine the fourth undetermined feature monitoring parameter and the corresponding fourth feature coefficient of the corresponding monitoring log set according to the third undetermined feature monitoring parameter and the corresponding third feature coefficient of the comparison combination of several monitoring log subsets of the same monitoring log set; The characteristic monitoring parameter and the corresponding characteristic coefficient are determined according to the second undetermined characteristic monitoring parameter, the fourth undetermined characteristic monitoring parameter and the corresponding second characteristic coefficient and the fourth characteristic coefficient.

[0010] In some embodiments of the present application, determining a second undetermined feature monitoring parameter and a corresponding second feature coefficient corresponding to a subset of monitoring logs includes: Mark the multi-source monitoring parameter sequence W of each relevant monitoring log in the monitoring stage, and calculate the first change value of each parameter type belonging to the same monitoring stage mark; If the first change value is greater than the preset parameter change value threshold and the historical life change value of the corresponding monitoring stage is greater than the preset life change value, set the corresponding historical monitoring parameter as the first pending monitoring parameter and calculate the corresponding first characteristic coefficient; Comparing a number of first undetermined feature monitoring parameters determined from different related monitoring logs of the same monitoring log subset to obtain a credibility coefficient for each first undetermined feature monitoring parameter, and eliminating first undetermined feature monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining first undetermined feature monitoring parameters are set as the second undetermined feature monitoring parameters of the corresponding monitoring log subset and the corresponding second feature coefficients are calculated.

[0011] In some embodiments of the present application, determining a fourth undetermined feature monitoring parameter and a corresponding fourth feature coefficient corresponding to a monitoring log set includes: Compare the historical total life values ​​of different monitoring log subsets of the same monitoring log set, and determine several monitoring log subset comparison combinations based on the comparison results; Wherein, each monitoring log subset comparison combination includes a target monitoring log subset and a plurality of monitoring log subsets; Randomly select a target monitoring log subset of the same monitoring log subset comparison combination and a related monitoring log of each monitoring log subset to construct a comparison subcombination of the monitoring log subset comparison combination; Constructing a number of comparison subcombinations of comparison combinations of each monitoring log subset; Pre-set several life analysis nodes; Obtain target-related monitoring logs and historical life values ​​of each related monitoring log in the same comparison subcombination according to the life analysis node, and generate a target life value change curve and several historical life value change curves; Compare several historical life value change curves with the target life value change curve, obtain the abnormal analysis node of each historical life value change curve, and calculate the abnormal life difference at the abnormal analysis node; Obtaining a second variation value of each historical monitoring parameter between an abnormality analysis node and a previous life analysis node in a corresponding related monitoring log, and screening out historical monitoring parameters whose second variation value is greater than a preset parameter variation value; Setting an arrangement order based on a number of abnormal life difference values ​​and generating an abnormal life difference value sequence; Arranging the selected second variation values ​​of the same historical monitoring parameter in an arrangement order, and constructing a second variation value sequence of each selected historical monitoring data; Calculate the correlation coefficient between the second variation value sequence and the corresponding abnormal life difference value sequence, and remove the historical monitoring parameters whose correlation coefficient is less than a preset correlation coefficient threshold; Comparing the remaining screened historical monitoring parameters of several comparison subcombinations of the same monitoring log subset comparison combination, determining the third undetermined characteristic monitoring parameter of the corresponding monitoring log subset comparison combination according to the comparison result, and calculating the corresponding third characteristic coefficient; Comparing a number of third undetermined characteristic monitoring parameters determined by comparing different monitoring log subsets of the same monitoring log set, obtaining a credibility coefficient for each third undetermined characteristic monitoring parameter, and eliminating third undetermined characteristic monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining third undetermined feature monitoring parameter is set as the fourth undetermined feature monitoring parameter of the corresponding monitoring log set and the corresponding fourth feature coefficient is calculated.

[0012] In some embodiments of the present application, characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors are determined based on the analysis results, including: Normalizing several environmental change factors involved in the same monitoring log set to obtain the characteristic environmental change factor of the corresponding monitoring log set; Configure corresponding compensation coefficients for the fourth undetermined characteristic monitoring parameters of the same monitoring log set and the second undetermined characteristic monitoring parameters determined for each monitoring log subset, and calculate the important coefficients of the same monitoring parameter based on the compensation coefficients; Setting the monitoring parameter whose importance coefficient is greater than the preset importance coefficient threshold as the characteristic monitoring parameter, and calculating the corresponding characteristic coefficient according to the corresponding second characteristic coefficient and the fourth characteristic coefficient; The characteristic environmental change factor of each monitoring log set is mapped with the corresponding characteristic monitoring parameter and the corresponding characteristic coefficient to obtain the characteristic monitoring parameter of the characteristic environmental change factor and the corresponding characteristic coefficient.

[0013] In some embodiments of the present application, setting a weight coefficient of a corresponding characteristic monitoring parameter according to the characteristic coefficient includes: The weight coefficient is calculated according to the number of characteristic monitoring parameters mapped to each characteristic environmental change factor and the characteristic coefficient of the corresponding characteristic monitoring parameter to obtain the weight coefficient of each characteristic monitoring parameter.

[0014] In some embodiments of the present application, constructing a lifespan prediction model includes: Obtaining a preset life value mapping table for each characteristic monitoring parameter, wherein the preset life value mapping table includes preset life values ​​mapped to a plurality of preset parameter values ​​of the characteristic monitoring parameter; The characteristic monitoring parameters at several preset collection nodes of each characteristic environmental change factor, the preset life values ​​mapped by the characteristic monitoring parameters, and the weight coefficients of the corresponding characteristic monitoring parameters are trained on a neural network to obtain a life prediction sub-model for each characteristic environmental change factor; The lifespan prediction sub-models of all characteristic environmental change factors are integrated to obtain the lifespan prediction model.

[0015] In some embodiments of the present application, generating a life prediction value based on a life prediction model, a real-time environmental change factor, and a real-time characteristic monitoring parameter, and determining whether to generate an alarm signal includes: Obtain real-time environmental parameters and calculate corresponding real-time environmental change factors; Perform similarity analysis on the real-time environmental change factor and multiple characteristic environmental change factors to obtain the characteristic environmental change factor with the greatest similarity to the real-time environmental change factor; Input the characteristic environmental change factors with the greatest similarity and the real-time characteristic monitoring parameters into the life prediction model to obtain the life prediction value; If the life prediction value is less than the preset life value threshold, an alarm signal is generated.

[0016] Compared with the prior art, the photovoltaic fuse life prediction method based on multi-parameter fusion in the embodiment of the present application has the following advantages: By determining several characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors, the characteristic monitoring parameters and corresponding weight coefficients under different photovoltaic environmental factors are obtained, and a life prediction model is constructed to determine the influence weights of multi-source monitoring parameters on the life value and perform parameter fusion to improve the accuracy of life prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a photovoltaic fuse life prediction method based on multi-parameter fusion in an embodiment of the present application. DETAILED DESCRIPTION

[0018] The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0019] In the description of this application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application.

[0020] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0021] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0022] like Figure 1 As shown, a photovoltaic fuse life prediction method based on multi-parameter fusion in an embodiment of the present application includes: Step S101: determining basic information of a photovoltaic fuse and obtaining relevant monitoring logs, obtaining and classifying photovoltaic environmental factors in each relevant monitoring log, and obtaining multiple monitoring log sets, wherein the monitoring log sets include several monitoring log subsets; Step S102: extracting and analyzing multi-source monitoring parameters of each relevant monitoring log in a plurality of monitoring log subsets, and determining characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors according to the analysis results; Step S103: setting a weight coefficient of a corresponding characteristic monitoring parameter according to the characteristic coefficient, and constructing a life prediction model according to several characteristic monitoring parameters of different photovoltaic environmental factors, corresponding weight coefficients, and several preset life values; Step S104: determining the real-time photovoltaic environment change factor, generating a life prediction value based on the life prediction model, the real-time environment change factor and the real-time characteristic monitoring parameter, and determining whether to generate an alarm signal.

[0023] In this embodiment, the basic information includes the manufacturing material, manufacturing process, model, etc. of the photovoltaic fuse, and the relevant monitoring log refers to the historical monitoring log of the photovoltaic fuse whose similarity with the basic information of the current photovoltaic fuse is greater than the preset similarity threshold.

[0024] In some embodiments of the present application, the photovoltaic environmental factors in each relevant monitoring log are obtained and classified to obtain multiple monitoring log sets, including: Obtain the historical total life value in each relevant monitoring log, and determine a number of monitoring stages corresponding to the relevant monitoring log according to the historical total life value and the preset monitoring interval; Pre-set a number of environmental monitoring and evaluation indicators; Obtain several historical environmental parameters for each monitoring phase in the same relevant monitoring log; Evaluate several historical environmental parameters based on several environmental monitoring and evaluation indicators, obtain reference evaluation values ​​of several environmental monitoring and evaluation indicators in each monitoring stage, and set them as photovoltaic environmental factors corresponding to relevant monitoring logs; The photovoltaic environmental factors of different relevant monitoring logs were analyzed for similarity, and similarity coefficients of different relevant monitoring logs were obtained; A monitoring log set is constructed based on related monitoring logs whose similarity coefficients are greater than a preset similarity coefficient threshold, and the same monitoring log set is further divided according to the corresponding total historical life values ​​to obtain several monitoring log subsets.

[0025] In this embodiment, the historical total life value refers to the service life of the photovoltaic fuse.

[0026] In this embodiment, the photovoltaic environmental factors include reference evaluation values ​​of several environmental monitoring evaluation indicators in all monitoring stages. Similarity analysis is performed on the photovoltaic environmental factors of different related monitoring logs to obtain similarity coefficients, thereby classifying the related monitoring logs and obtaining a set of monitoring logs with similar photovoltaic environmental factors, laying the foundation for the subsequent construction of a life prediction model.

[0027] In this embodiment, the environmental monitoring evaluation indicators include the change rate, change trend, and fluctuation degree of historical environmental parameters, which comprehensively evaluate the consistency of environmental changes in the corresponding monitoring stages of different relevant monitoring logs, lay the foundation for the subsequent construction of a life prediction model and life prediction, and improve the accuracy of life prediction.

[0028] In this embodiment, the monitoring log set refers to several related monitoring logs with relatively similar photovoltaic environmental factors, and the monitoring log subset refers to several related monitoring logs with relatively small differences in historical total life values ​​in the same monitoring log set.

[0029] In this embodiment, different historical total life values ​​correspond to different preset monitoring intervals. When the historical total life value is larger, the corresponding preset monitoring interval is larger, and vice versa. The relevant monitoring logs are divided into multiple monitoring stages according to the preset time intervals to improve the monitoring and evaluation accuracy of photovoltaic environmental factors and lay the foundation for the subsequent calculation of the similarity coefficient.

[0030] In some embodiments of the present application, obtaining similarity coefficients of different related monitoring logs includes: Randomly select a relevant monitoring log and set it as the target relevant monitoring log; Compare the photovoltaic environmental factors of each monitoring stage of the target-related monitoring log with the photovoltaic environmental factors of the corresponding monitoring stage of each related monitoring log, and obtain the sub-similarity coefficients of each monitoring stage of the target-related monitoring log and the corresponding monitoring stage of the corresponding related monitoring log; The calculation formula of the sub-similarity coefficient is: ; Among them, P1 is the sub-similarity coefficient, p0 is the similarity conversion coefficient, and n1 is the total number of environmental monitoring and evaluation indicators. is the reference evaluation value of the i1th environmental monitoring evaluation indicator of the target-related monitoring log in the current monitoring phase, is the reference evaluation value of the i1th environmental monitoring evaluation indicator in the corresponding monitoring log in the corresponding monitoring stage, and ai is the weight coefficient of the i-th environmental monitoring evaluation indicator; Generate a similarity coefficient between the target-related monitoring log and the corresponding related monitoring log according to several sub-similarity coefficients of all monitoring stages of the target-related monitoring log and the corresponding monitoring stages of the same related monitoring log; The calculation formula of the similarity coefficient is: ; Among them, P2 is the similarity coefficient, is the sub-similarity coefficient of the target related monitoring log and the corresponding related monitoring log in the i2th monitoring stage, is the sub-similarity coefficient threshold, is the weight coefficient of the i2th monitoring stage of the target-related monitoring log, and n2 is the total number of monitoring stages of the target-related monitoring log; Generate the similarity coefficient between the target related monitoring log and each related monitoring log in turn.

[0031] In some embodiments of the present application, extracting and analyzing multi-source monitoring parameters of each relevant monitoring log in several monitoring log subsets includes: Obtain the historical life span change value of each monitoring stage of each relevant monitoring log in the monitoring log subset, and set a number of extraction time nodes of the corresponding monitoring stage according to the historical life span change value and the corresponding preset time interval; Extract the multi-source monitoring parameters of the corresponding monitoring stage of each relevant monitoring log according to the extraction time node to obtain the multi-source monitoring parameter matrix W of each relevant monitoring log;

[0032] in, The historical monitoring parameters at the s1th extraction time node of the s2th parameter type of the relevant monitoring log, s1=1,…r, s2=1,…m; Analyze the multi-source monitoring parameter sequence W of each relevant monitoring log to determine a number of first undetermined characteristic monitoring parameters and corresponding first characteristic coefficients corresponding to the relevant monitoring log; Determine a second undetermined feature monitoring parameter and a corresponding second feature coefficient of a corresponding monitoring log subset according to a plurality of first undetermined feature monitoring parameters and corresponding first feature coefficients determined by different related monitoring logs of the same monitoring log subset; Constructing a plurality of monitoring log subset comparison combinations of the same monitoring log set and a plurality of comparison subcombinations of each monitoring log subset comparison combination; Analyze several comparison subcombinations to determine the third undetermined characteristic monitoring parameter and the corresponding third characteristic coefficient of each monitoring log subset comparison combination; Determine the fourth undetermined feature monitoring parameter and the corresponding fourth feature coefficient of the corresponding monitoring log set according to the third undetermined feature monitoring parameter and the corresponding third feature coefficient of the comparison combination of several monitoring log subsets of the same monitoring log set; The characteristic monitoring parameter and the corresponding characteristic coefficient are determined according to the second undetermined characteristic monitoring parameter, the fourth undetermined characteristic monitoring parameter and the corresponding second characteristic coefficient and the fourth characteristic coefficient.

[0033] In this embodiment, when the historical life span variation value is larger and the corresponding preset time interval is smaller, more extraction time nodes are set for the corresponding monitoring stage, and vice versa.

[0034] In some embodiments of the present application, determining a second undetermined feature monitoring parameter and a corresponding second feature coefficient corresponding to a subset of monitoring logs includes: Mark the multi-source monitoring parameter sequence W of each relevant monitoring log in the monitoring stage, and calculate the first change value of each parameter type belonging to the same monitoring stage mark; If the first change value is greater than the preset parameter change value threshold and the historical life change value of the corresponding monitoring stage is greater than the preset life change value, set the corresponding historical monitoring parameter as the first pending monitoring parameter and calculate the corresponding first characteristic coefficient; Comparing a number of first undetermined feature monitoring parameters determined from different related monitoring logs of the same monitoring log subset to obtain a credibility coefficient for each first undetermined feature monitoring parameter, and eliminating first undetermined feature monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining first undetermined feature monitoring parameters are set as the second undetermined feature monitoring parameters of the corresponding monitoring log subset and the corresponding second feature coefficients are calculated.

[0035] In this embodiment, the first characteristic coefficient = characteristic conversion coefficient * (historical life change value in the monitoring phase / first change value of the corresponding first pending monitoring parameter in the monitoring phase). When the ratio is larger, the first characteristic coefficient is larger, and vice versa.

[0036] In this embodiment, the second characteristic coefficient is calculated based on multiple first characteristic coefficients of the remaining first undetermined characteristic monitoring parameters and corresponding credibility coefficients.

[0037] In this embodiment, by analyzing the historical life change values ​​of each monitoring stage of the same monitoring log subset and the first change values ​​of the historical monitoring parameters, the second pending monitoring parameters of the monitoring log subset are obtained, which lays the foundation for the subsequent determination of characteristic monitoring parameters and the construction of a life prediction model, thereby improving the accuracy of photovoltaic fuse life prediction.

[0038] In some embodiments of the present application, determining a fourth undetermined feature monitoring parameter and a corresponding fourth feature coefficient corresponding to a monitoring log set includes: Compare the historical total life values ​​of different monitoring log subsets of the same monitoring log set, and determine several monitoring log subset comparison combinations based on the comparison results; Wherein, each monitoring log subset comparison combination includes a target monitoring log subset and a plurality of monitoring log subsets; Randomly select a target monitoring log subset of the same monitoring log subset comparison combination and a related monitoring log of each monitoring log subset to construct a comparison subcombination of the monitoring log subset comparison combination; Constructing a number of comparison subcombinations of comparison combinations of each monitoring log subset; Pre-set several life analysis nodes; Obtain target-related monitoring logs and historical life values ​​of each related monitoring log in the same comparison subcombination according to the life analysis node, and generate a target life value change curve and several historical life value change curves; Compare several historical life value change curves with the target life value change curve, obtain the abnormal analysis node of each historical life value change curve, and calculate the abnormal life difference at the abnormal analysis node; Obtaining a second variation value of each historical monitoring parameter between an abnormality analysis node and a previous life analysis node in a corresponding related monitoring log, and screening out historical monitoring parameters whose second variation value is greater than a preset parameter variation value; Setting an arrangement order based on a number of abnormal life difference values ​​and generating an abnormal life difference value sequence; Arranging the selected second variation values ​​of the same historical monitoring parameter in an arrangement order, and constructing a second variation value sequence of each selected historical monitoring data; Calculate the correlation coefficient between the second variation value sequence and the corresponding abnormal life difference value sequence, and remove the historical monitoring parameters whose correlation coefficient is less than a preset correlation coefficient threshold; Comparing the remaining screened historical monitoring parameters of several comparison subcombinations of the same monitoring log subset comparison combination, determining the third undetermined characteristic monitoring parameter of the corresponding monitoring log subset comparison combination according to the comparison result, and calculating the corresponding third characteristic coefficient; Comparing a number of third undetermined characteristic monitoring parameters determined by comparing different monitoring log subsets of the same monitoring log set, obtaining a credibility coefficient for each third undetermined characteristic monitoring parameter, and eliminating third undetermined characteristic monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining third undetermined feature monitoring parameter is set as the fourth undetermined feature monitoring parameter of the corresponding monitoring log set and the corresponding fourth feature coefficient is calculated.

[0039] In this embodiment, the monitoring log subset comparison combination refers to several monitoring log subsets in the same monitoring log set whose historical total life value difference with the target monitoring log subset is greater than the preset total life value difference, and the number of monitoring log subsets is at least one.

[0040] In this embodiment, the abnormal analysis node refers to the initial analysis node where the historical life value change curve and the target life value change curve have large differences in historical life values ​​at the same life analysis node, and the abnormal life difference refers to the life difference between the historical life value in the historical life value change curve at the abnormal analysis node and the historical life value in the target life value change curve.

[0041] In this embodiment, the third undetermined characteristic monitoring parameter refers to the credibility coefficient of the remaining screened historical monitoring parameters of several comparison sub-combinations, that is, the remaining screened historical monitoring parameters whose credibility coefficients are greater than a preset credibility coefficient threshold.

[0042] In this embodiment, the correlation coefficient refers to the influence of the second change value in the second change value sequence on the corresponding abnormal life difference in the abnormal life difference sequence. When both have an influence and the greater the influence, the larger the corresponding correlation coefficient, and vice versa.

[0043] In this embodiment, the third characteristic coefficient is determined based on the fifth characteristic coefficient and the corresponding credibility coefficient corresponding to the remaining screened historical monitoring parameters. The fifth characteristic coefficient corresponding to the remaining screened historical monitoring parameters refers to the ratio of several abnormal life differences to the corresponding second change values ​​and the correlation coefficient. When the ratio is larger and the correlation coefficient is larger, the fifth characteristic coefficient corresponding to the remaining screened historical monitoring parameters is larger, and in any case, the smaller it is.

[0044] In this embodiment, by analyzing the historical monitoring parameters of related monitoring logs with different historical total life values ​​and similar photovoltaic environmental factors, several fourth pending characteristic monitoring parameters of the corresponding monitoring log set are obtained, which lays the foundation for the subsequent determination of characteristic monitoring parameters and construction of a life prediction model, thereby improving the accuracy of photovoltaic fuse life prediction.

[0045] In some embodiments of the present application, characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors are determined based on the analysis results, including: Normalizing several environmental change factors involved in the same monitoring log set to obtain the characteristic environmental change factor of the corresponding monitoring log set; Configure corresponding compensation coefficients for the fourth undetermined characteristic monitoring parameters of the same monitoring log set and the second undetermined characteristic monitoring parameters determined for each monitoring log subset, and calculate the important coefficients of the same monitoring parameter based on the compensation coefficients; Setting the monitoring parameter whose importance coefficient is greater than the preset importance coefficient threshold as the characteristic monitoring parameter, and calculating the corresponding characteristic coefficient according to the corresponding second characteristic coefficient and the fourth characteristic coefficient; The characteristic environmental change factor of each monitoring log set is mapped with the corresponding characteristic monitoring parameter and the corresponding characteristic coefficient to obtain the characteristic monitoring parameter of the characteristic environmental change factor and the corresponding characteristic coefficient.

[0046] In this embodiment, the compensation coefficient configured for each fourth pending characteristic monitoring parameter is 0.4, and the compensation coefficient configured for the second pending characteristic monitoring parameter determined by each monitoring log subset is 0.2. If the same monitoring parameter appears in several fourth pending characteristic monitoring parameters and also appears in several second pending characteristic monitoring parameters determined by several monitoring log subsets, the calculation formula of the importance coefficient is = 0.4 + 0.2 * the number of appearances in several monitoring log subsets. The more important coefficients there are, the greater the impact of the corresponding monitoring parameter on the life of the photovoltaic fuse, that is, the greater the impact on the subsequent life prediction.

[0047] In this embodiment, the characteristic environmental change factor refers to the average of reference evaluation values ​​of several environmental monitoring evaluation indicators corresponding to the monitoring log set.

[0048] In some embodiments of the present application, setting a weight coefficient of a corresponding characteristic monitoring parameter according to the characteristic coefficient includes: The weight coefficient is calculated according to the number of characteristic monitoring parameters mapped to each characteristic environmental change factor and the characteristic coefficient of the corresponding characteristic monitoring parameter to obtain the weight coefficient of each characteristic monitoring parameter.

[0049] In some embodiments of the present application, constructing a lifespan prediction model includes: Obtaining a preset life value mapping table for each characteristic monitoring parameter, wherein the preset life value mapping table includes preset life values ​​mapped to a plurality of preset parameter values ​​of the characteristic monitoring parameter; The characteristic monitoring parameters at several preset collection nodes of each characteristic environmental change factor, the preset life values ​​mapped by the characteristic monitoring parameters, and the weight coefficients of the corresponding characteristic monitoring parameters are trained on a neural network to obtain a life prediction sub-model for each characteristic environmental change factor; The lifespan prediction sub-models of all characteristic environmental change factors are integrated to obtain the lifespan prediction model.

[0050] In this embodiment, the preset life value mapping table is constructed based on the life values ​​corresponding to the historical monitoring parameters when the life assessment is performed on a single parameter type. The preset life value refers to the remaining life of the photovoltaic fuse.

[0051] In this embodiment, by constructing a life prediction sub-model for each characteristic environmental change factor and fusing the life prediction sub-models of different characteristic environmental change factors for training, the life prediction model can be improved to provide more accurate life prediction values ​​under different environmental changes and characteristic monitoring parameters with different weights.

[0052] In some embodiments of the present application, generating a life prediction value based on a life prediction model, a real-time environmental change factor, and a real-time characteristic monitoring parameter, and determining whether to generate an alarm signal includes: Obtain real-time environmental parameters and calculate corresponding real-time environmental change factors; Perform similarity analysis on the real-time environmental change factor and multiple characteristic environmental change factors to obtain the characteristic environmental change factor with the greatest similarity to the real-time environmental change factor; Input the characteristic environmental change factors with the greatest similarity and the real-time characteristic monitoring parameters into the life prediction model to obtain the life prediction value; If the life prediction value is less than the preset life value threshold, an alarm signal is generated.

[0053] In this embodiment, the life prediction value refers to the predicted remaining life of the photovoltaic fuse. The characteristic environmental change factor with the greatest similarity is determined through the real-time environmental change factor, thereby determining the weight coefficient of the real-time characteristic monitoring parameter, and accurately obtaining the life prediction value, thereby improving the accuracy of the predicted remaining life of the photovoltaic fuse.

[0054] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and replacements can be made without departing from the technical principles of the present application. These improvements and replacements should also be regarded as the scope of protection of the present application.

Claims

1. A photovoltaic fuse life prediction method based on multi-parameter fusion, characterized in that: include: Determine basic information of the photovoltaic fuse and obtain relevant monitoring logs, obtain and classify photovoltaic environmental factors in each relevant monitoring log, and obtain multiple monitoring log sets, wherein the monitoring log sets include several monitoring log subsets; Extract and analyze the multi-source monitoring parameters of each relevant monitoring log in several monitoring log subsets, and determine the characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors based on the analysis results; The weight coefficient of the corresponding characteristic monitoring parameter is set according to the characteristic coefficient, and the life prediction model is constructed according to several characteristic monitoring parameters of different photovoltaic environmental factors, the corresponding weight coefficient and several preset life values; Determine the real-time photovoltaic environment change factor, generate a life prediction value based on the life prediction model, the real-time environment change factor and the real-time characteristic monitoring parameters, and determine whether to generate an alarm signal.

2. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 1, characterized in that: The photovoltaic environmental factors in each relevant monitoring log are obtained and classified to obtain multiple monitoring log sets, including: Obtain the historical total life value in each relevant monitoring log, and determine a number of monitoring stages corresponding to the relevant monitoring log according to the historical total life value and the preset monitoring interval; Pre-set a number of environmental monitoring and evaluation indicators; Obtain several historical environmental parameters for each monitoring phase in the same relevant monitoring log; Evaluate several historical environmental parameters based on several environmental monitoring and evaluation indicators, obtain reference evaluation values ​​of several environmental monitoring and evaluation indicators in each monitoring stage, and set them as photovoltaic environmental factors corresponding to relevant monitoring logs; The photovoltaic environmental factors of different relevant monitoring logs were analyzed for similarity, and similarity coefficients of different relevant monitoring logs were obtained; A monitoring log set is constructed based on related monitoring logs whose similarity coefficients are greater than a preset similarity coefficient threshold, and the same monitoring log set is further divided according to the corresponding total historical life values ​​to obtain several monitoring log subsets.

3. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 2, characterized in that: Get the similarity coefficients of different related monitoring logs, including: Randomly select a relevant monitoring log and set it as the target relevant monitoring log; Compare the photovoltaic environmental factors of each monitoring stage of the target-related monitoring log with the photovoltaic environmental factors of the corresponding monitoring stage of each related monitoring log, and obtain the sub-similarity coefficients of each monitoring stage of the target-related monitoring log and the corresponding monitoring stage of the corresponding related monitoring log; The calculation formula of the sub-similarity coefficient is: ； Among them, P1 is the sub-similarity coefficient, p0 is the similarity conversion coefficient, and n1 is the total number of environmental monitoring and evaluation indicators. is the reference evaluation value of the i1th environmental monitoring evaluation indicator of the target-related monitoring log in the current monitoring phase, is the reference evaluation value of the i1th environmental monitoring evaluation indicator in the corresponding monitoring log in the corresponding monitoring stage, and ai is the weight coefficient of the i-th environmental monitoring evaluation indicator; Generate a similarity coefficient between the target-related monitoring log and the corresponding related monitoring log according to several sub-similarity coefficients of all monitoring stages of the target-related monitoring log and the corresponding monitoring stages of the same related monitoring log; The calculation formula of the similarity coefficient is: ； Among them, P2 is the similarity coefficient, is the sub-similarity coefficient of the target related monitoring log and the corresponding related monitoring log in the i2th monitoring stage, is the sub-similarity coefficient threshold, is the weight coefficient of the i2th monitoring stage of the target-related monitoring log, and n2 is the total number of monitoring stages of the target-related monitoring log; Generate the similarity coefficient between the target related monitoring log and each related monitoring log in turn.

4. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 3, characterized in that: Extract and analyze the multi-source monitoring parameters of each relevant monitoring log in several monitoring log subsets, including: Obtain the historical life span change value of each monitoring stage of each relevant monitoring log in the monitoring log subset, and set a number of extraction time nodes of the corresponding monitoring stage according to the historical life span change value and the corresponding preset time interval; Extract the multi-source monitoring parameters of the corresponding monitoring stage of each relevant monitoring log according to the extraction time node to obtain the multi-source monitoring parameter matrix W of each relevant monitoring log; in, The historical monitoring parameters at the s1th extraction time node of the s2th parameter type of the relevant monitoring log, s1=1,…r, s2=1,…m; Analyze the multi-source monitoring parameter sequence W of each relevant monitoring log to determine a number of first undetermined characteristic monitoring parameters and corresponding first characteristic coefficients corresponding to the relevant monitoring log; Determine a second undetermined feature monitoring parameter and a corresponding second feature coefficient of a corresponding monitoring log subset according to a plurality of first undetermined feature monitoring parameters and corresponding first feature coefficients determined by different related monitoring logs of the same monitoring log subset; Constructing a plurality of monitoring log subset comparison combinations of the same monitoring log set and a plurality of comparison subcombinations of each monitoring log subset comparison combination; Analyze several comparison subcombinations to determine the third undetermined characteristic monitoring parameter and the corresponding third characteristic coefficient of each monitoring log subset comparison combination; Determine the fourth undetermined feature monitoring parameter and the corresponding fourth feature coefficient of the corresponding monitoring log set according to the third undetermined feature monitoring parameter and the corresponding third feature coefficient of the comparison combination of several monitoring log subsets of the same monitoring log set; The characteristic monitoring parameter and the corresponding characteristic coefficient are determined according to the second undetermined characteristic monitoring parameter, the fourth undetermined characteristic monitoring parameter and the corresponding second characteristic coefficient and the fourth characteristic coefficient.

5. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 4, characterized in that: Determining a second undetermined characteristic monitoring parameter corresponding to the monitoring log subset and a corresponding second characteristic coefficient includes: Mark the multi-source monitoring parameter sequence W of each relevant monitoring log in the monitoring stage, and calculate the first change value of each parameter type belonging to the same monitoring stage mark; If the first change value is greater than the preset parameter change value threshold and the historical life change value of the corresponding monitoring stage is greater than the preset life change value, set the corresponding historical monitoring parameter as the first pending monitoring parameter and calculate the corresponding first characteristic coefficient; Comparing a number of first undetermined feature monitoring parameters determined from different related monitoring logs of the same monitoring log subset to obtain a credibility coefficient for each first undetermined feature monitoring parameter, and eliminating first undetermined feature monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining first undetermined feature monitoring parameters are set as the second undetermined feature monitoring parameters of the corresponding monitoring log subset and the corresponding second feature coefficients are calculated.

6. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 5, characterized in that: Determining a fourth undetermined characteristic monitoring parameter and a corresponding fourth characteristic coefficient corresponding to the monitoring log set includes: Compare the historical total life values ​​of different monitoring log subsets of the same monitoring log set, and determine several monitoring log subset comparison combinations based on the comparison results; Wherein, each monitoring log subset comparison combination includes a target monitoring log subset and a plurality of monitoring log subsets; Randomly select a target monitoring log subset of the same monitoring log subset comparison combination and a related monitoring log of each monitoring log subset to construct a comparison subcombination of the monitoring log subset comparison combination; Constructing a number of comparison subcombinations of comparison combinations of each monitoring log subset; Pre-set several life analysis nodes; Obtain target-related monitoring logs and historical life values ​​of each related monitoring log in the same comparison subcombination according to the life analysis node, and generate a target life value change curve and several historical life value change curves; Compare several historical life value change curves with the target life value change curve, obtain the abnormal analysis node of each historical life value change curve, and calculate the abnormal life difference at the abnormal analysis node; Obtaining a second variation value of each historical monitoring parameter between an abnormality analysis node and a previous life analysis node in a corresponding related monitoring log, and screening out historical monitoring parameters whose second variation value is greater than a preset parameter variation value; Setting an arrangement order based on a number of abnormal life difference values ​​and generating an abnormal life difference value sequence; Arranging the selected second variation values ​​of the same historical monitoring parameter in an arrangement order, and constructing a second variation value sequence of each selected historical monitoring data; Calculate the correlation coefficient between the second variation value sequence and the corresponding abnormal life difference value sequence, and remove the historical monitoring parameters whose correlation coefficient is less than a preset correlation coefficient threshold; Comparing the remaining screened historical monitoring parameters of several comparison subcombinations of the same monitoring log subset comparison combination, determining the third undetermined characteristic monitoring parameter of the corresponding monitoring log subset comparison combination according to the comparison result, and calculating the corresponding third characteristic coefficient; Comparing a number of third undetermined characteristic monitoring parameters determined by comparing different monitoring log subsets of the same monitoring log set, obtaining a credibility coefficient for each third undetermined characteristic monitoring parameter, and eliminating third undetermined characteristic monitoring parameters having a credibility coefficient less than a preset credibility coefficient threshold; The remaining third undetermined feature monitoring parameter is set as the fourth undetermined feature monitoring parameter of the corresponding monitoring log set and the corresponding fourth feature coefficient is calculated.

7. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 6, characterized in that: Based on the analysis results, the characteristic monitoring parameters and corresponding characteristic coefficients of different characteristic photovoltaic environmental factors are determined, including: Normalizing several environmental change factors involved in the same monitoring log set to obtain the characteristic environmental change factor of the corresponding monitoring log set; Configure corresponding compensation coefficients for the fourth undetermined characteristic monitoring parameters of the same monitoring log set and the second undetermined characteristic monitoring parameters determined for each monitoring log subset, and calculate the important coefficients of the same monitoring parameter based on the compensation coefficients; Setting the monitoring parameter whose importance coefficient is greater than the preset importance coefficient threshold as the characteristic monitoring parameter, and calculating the corresponding characteristic coefficient according to the corresponding second characteristic coefficient and the fourth characteristic coefficient; The characteristic environmental change factor of each monitoring log set is mapped with the corresponding characteristic monitoring parameter and the corresponding characteristic coefficient to obtain the characteristic monitoring parameter of the characteristic environmental change factor and the corresponding characteristic coefficient.

8. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 7, characterized in that: The weight coefficients of the corresponding characteristic monitoring parameters are set according to the characteristic coefficients, including: The weight coefficient is calculated according to the number of characteristic monitoring parameters mapped to each characteristic environmental change factor and the characteristic coefficient of the corresponding characteristic monitoring parameter to obtain the weight coefficient of each characteristic monitoring parameter.

9. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 8, characterized in that: Construct a lifespan prediction model, including: Obtaining a preset life value mapping table for each characteristic monitoring parameter, wherein the preset life value mapping table includes preset life values ​​mapped to a plurality of preset parameter values ​​of the characteristic monitoring parameter; The characteristic monitoring parameters at several preset collection nodes of each characteristic environmental change factor, the preset life values ​​mapped by the characteristic monitoring parameters, and the weight coefficients of the corresponding characteristic monitoring parameters are trained on a neural network to obtain a life prediction sub-model for each characteristic environmental change factor; The lifespan prediction sub-models of all characteristic environmental change factors are integrated to obtain the lifespan prediction model.

10. The photovoltaic fuse life prediction method based on multi-parameter fusion according to claim 9, characterized in that: Generate life prediction values ​​based on the life prediction model, real-time environmental change factors, and real-time characteristic monitoring parameters, and determine whether to generate an alarm signal, including: Obtain real-time environmental parameters and calculate corresponding real-time environmental change factors; Perform similarity analysis on the real-time environmental change factor and multiple characteristic environmental change factors to obtain the characteristic environmental change factor with the greatest similarity to the real-time environmental change factor; Input the characteristic environmental change factors with the greatest similarity and the real-time characteristic monitoring parameters into the life prediction model to obtain the life prediction value; If the life prediction value is less than the preset life value threshold, an alarm signal is generated.

Images