Power generation anomaly early warning method and system for photovoltaic power generation system

Abstract

The application relates to the technical field of photovoltaic power generation, in particular to a power generation anomaly early warning method and system for a photovoltaic power generation system. The method comprises the following steps: constructing a power fluctuation graph; performing threshold segmentation and clustering processing on the power fluctuation graph to obtain an anomaly cluster at a current moment; obtaining a union set of any anomaly cluster within continuous multiple historical moments; taking the similarity of an anomaly sequence as a weight to calculate the weighted correlation of the output power between any cell panel and other cell panels, and obtaining the spatial correlation of each anomaly cell panel; dividing the anomaly cell panel into a real fault and a false fault according to the spatial correlation and the change rate of the power fluctuation, and then issuing an anomaly early warning. Through the technical scheme, the power generation anomaly caused by environmental interference can be effectively distinguished, and the accuracy of the early warning result is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation, in particular to a power generation anomaly early warning method and system for a photovoltaic power generation system. BACKGROUND

[0002] A photovoltaic power generation system refers to a power generation control system that converts solar radiation energy into electric energy directly through the photovoltaic effect of a photovoltaic cell. During the power generation operation of the photovoltaic power generation system, its abnormal state needs to be monitored so as to facilitate timely removal of related faults, thereby facilitating subsequent maintenance.

[0003] At present, a patent application file with the application publication number CN117318616A discloses a photovoltaic early warning method, system, storage medium and electronic device. The method therein includes: when a fluctuation anomaly of a first signal collected by a first device is monitored, determining a target photovoltaic module corresponding to the first signal, the fluctuation anomaly being that the amplitude fluctuation of the first signal exceeds a first threshold value; controlling a second device to collect a second signal in a target area, the target area being a surrounding area centered on the target photovoltaic module; the second device includes a third device and a fourth device, the second signal includes a third signal and a fourth signal; the control of the second device to collect the second signal in the target area includes: controlling the third device to collect the third signal of the target photovoltaic module; controlling the fourth device to collect the fourth signal of other photovoltaic modules in the target area except the target photovoltaic module, the fourth signal being a signal of the same type as the first signal; if a change anomaly occurs in the second signal, outputting a corresponding early warning signal based on the fluctuation anomaly of the first signal, the change anomaly being that the amplitude change of the second signal exceeds a second threshold value.

[0004] The above method can determine whether the fluctuation of the first signal is noise interference by judging whether the second signal collected by the second device has a change anomaly when the fluctuation anomaly of the first signal is monitored. However, the photovoltaic power generation system is greatly affected by the environment, and the above method cannot effectively distinguish the fluctuation of the first signal caused by environmental interference, resulting in inaccurate early warning results. SUMMARY

[0005] In order to solve the technical problem of inaccurate power generation anomaly early warning of the photovoltaic power generation system, the present application provides a power generation anomaly early warning method and system for a photovoltaic power generation system, which can effectively distinguish the power generation anomaly caused by environmental interference and improve the accuracy of early warning results.

[0006] In a first aspect, this application provides a method for early warning of power generation anomalies in a photovoltaic power generation system. The method includes: calculating the absolute value of the difference between the output power of any solar panel and the average output power of each adjacent solar panel to obtain the power fluctuation of the solar panel, wherein the power fluctuations of each solar panel constitute a power fluctuation map of the photovoltaic power generation system; performing threshold segmentation on the power fluctuation map to obtain the abnormal solar panels at the current moment, and clustering the abnormal solar panels to obtain the abnormal clusters at the current moment; obtaining the union of any abnormal clusters over multiple consecutive historical moments, and calculating the similarity of the abnormal sequences between any solar panel and other solar panels within the union, using the similarity of the abnormal sequences as weights to calculate the weighted correlation of the output power between any solar panel and other solar panels to obtain the spatial correlation of each abnormal solar panel at the current moment, wherein the abnormal sequence includes a category label indicating whether the solar panel belongs to the abnormal solar panel over multiple consecutive historical moments; classifying the abnormal solar panels at the current moment into real faults and pseudo faults caused by environmental changes based on the spatial correlation of the abnormal solar panels and the rate of change of power fluctuations over multiple consecutive historical moments, thereby issuing an anomaly warning.

[0007] By constructing a power fluctuation map to initially identify anomalies, and combining the spatial correlation and power fluctuation change rate obtained from time series analysis, the system can effectively distinguish between real faults and false faults caused by environmental factors such as shading, significantly improving the accuracy of photovoltaic system anomaly early warning and reducing the false alarm rate.

[0008] Preferably, obtaining the union of any abnormal clusters includes: taking historical moments In the present moment The historical anomaly cluster with the largest intersection among the anomaly clusters is selected as the target cluster. Obtain the union of the target cluster and the abnormal cluster; then re-convert the historical time... The historical anomaly cluster with the largest intersection with the union of the given set is taken as the target cluster. The intersection is updated until the union of the abnormal clusters over multiple consecutive historical time periods is obtained.

[0009] By tracking the historical cluster with the largest intersection with the current anomaly cluster in the time series, it is ensured that the solar panels in the union are affected by the same moving environmental factor, accurately capturing the complete area affected by the entire environmental disturbance event, providing a data foundation for subsequent calculation of spatial correlation, and eliminating errors in spatial correlation caused by different environmental factors.

[0010] Preferably, the similarity of the anomalous sequences is negatively correlated with the Hamming distance of the anomalous sequences.

[0011] Preferably, the adjacent solar panel refers to any solar panel other than the solar panel itself within a circular area of ​​a predetermined radius centered on the solar panel.

[0012] It effectively eliminates macroscopic differences in environmental factors such as sunlight and temperature caused by excessive spatial distance in photovoltaic power generation systems, enabling power fluctuations to reflect the state differences between the target solar panel and its adjacent units, thereby more accurately highlighting the local anomalies of each solar panel in the photovoltaic power generation system.

[0013] Preferably, threshold segmentation of the power fluctuation map includes: determining a segmentation threshold for the power fluctuation map using the maximum inter-class variance method, and marking solar panels with power fluctuations greater than the segmentation threshold as abnormal solar panels.

[0014] Preferably, the solar panel Weighted correlation of output power with other solar panels Satisfying the relation:

[0015] ;

[0016] in, For the solar panels in the union Compared with other solar panels The similarity between anomalous sequences, The number of other solar panels in the union. For solar panels The sum of similarities between the anomalous sequences and those of other solar panels. For solar panels Compared with other solar panels The Pearson correlation coefficient between the output power.

[0017] By using As a weighting factor, weighted correlation focuses more on the correlation between solar panels with consistent time-series patterns, eliminating the reduction in output power correlation caused by differences in anomalous sequences within the union, and accurately measuring the correlation between solar panels. Correlation with other solar panels in the union in terms of output power.

[0018] Preferably, the abnormal solar panels are clustered to obtain the abnormal clusters at the current time, including: initializing the neighborhood radius and the minimum number of samples, and using the DBSCAN algorithm to cluster the abnormal solar panels to obtain multiple abnormal clusters.

[0019] Preferably, classifying the abnormal solar panel at the current moment into a real fault and a pseudo fault caused by environmental changes based on the spatial correlation and the rate of change of power fluctuation of the abnormal solar panel includes: in response to the spatial correlation being less than the correlation threshold and the rate of change of power fluctuation being less than the rate of change threshold, determining the abnormal solar panel as a real fault and issuing an abnormal warning; otherwise, determining the abnormal solar panel as a pseudo fault caused by environmental changes.

[0020] Preferably, when the battery panel belongs to an abnormal battery panel, the category label of the battery panel is 1, and vice versa, the category label of the battery panel is 0.

[0021] The category label of the battery panel is defined, and then an abnormal sequence is constructed, so that the abnormal sequence can accurately quantify the abnormal state of each battery panel in the set over time, and provide a data basis for subsequent calculation of weighted correlation.

[0022] The second aspect of the application also provides a power generation anomaly early warning system for a photovoltaic power generation system, comprising a processor and a memory, the memory storing computer program instructions, when the computer program instructions are executed by the processor, realizing the power generation anomaly early warning method for the photovoltaic power generation system according to the first aspect of the application.

[0023] The technical solution of the application has the following beneficial technical effects:

[0024] Firstly, by calculating the difference between the output power average value of any battery panel and its adjacent battery panel, a power fluctuation map reflecting the running state of each battery panel of the entire photovoltaic system is constructed, and threshold segmentation and DBSCAN clustering algorithm are used to automatically identify and divide the spatial distribution of abnormal battery panels at the current time from the power fluctuation map; further, in order to distinguish between real faults and pseudo faults caused by environmental interference, the abnormal clusters are tracked at continuous multiple historical times, and the battery panel area affected by the same environmental factor is determined by calculating the union. In this area, the similarity of the abnormal sequence of each battery panel is taken as the weight, and the weighted correlation of the output power is calculated, so as to obtain the spatial correlation representing the consistency of the output power change; finally, the spatial correlation and the change rate of the abnormal battery panel itself in the time sequence are comprehensively considered, to determine whether the abnormality is a real fault or a pseudo fault, and only the real fault is warned, so as to effectively improve the accuracy of the warning. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a flowchart of the power generation anomaly early warning method for a photovoltaic power generation system according to an embodiment of the application.

[0026] Figure 2 is a schematic diagram of a power fluctuation map according to an embodiment of the application.

[0027] Figure 3 is the identification result of real faults and pseudo faults caused by environmental changes according to an embodiment of the application.

[0028] Figure 4 is a structural block diagram of the power generation anomaly early warning system for a photovoltaic power generation system according to an embodiment of the application. DETAILED DESCRIPTION

[0029] According to a first aspect of the present application, the present application provides a power generation abnormality early warning method for a photovoltaic power generation system. The photovoltaic power generation system comprises a plurality of panels arranged in an array, each panel is adjacent to each other, and the output power of each panel can be collected, and the output power can reflect the power generation condition of each panel at each moment.

[0030] Figure 1 is a flowchart of the power generation abnormality early warning method for a photovoltaic power generation system according to an embodiment of the present application. As shown in Figure 1 , the power generation abnormality early warning method for a photovoltaic power generation system comprises steps S101 to S104, which are described in detail below.

[0031] S101, calculate the absolute value of the difference between the output power of any panel and the average value of the output power of each adjacent panel, to obtain the power fluctuation of the panel, and the power fluctuation of each panel constitutes a power fluctuation map of the photovoltaic power generation system.

[0032] In one embodiment, each panel in the photovoltaic power generation system corresponds to a coordinate position, the output power of any panel at the current moment is collected, and the adjacent panel is the panel in the circular region with the panel as the center and a predetermined radius, except the panel. The power fluctuation of the panel satisfies the relationship:

[0033] ;

[0034] wherein, is the value of the output power of the panel at the current moment, is a set composed of adjacent panels of the panel, is the value of the output power of the adjacent panel, is the number of adjacent panels in the set .

[0035] It should be noted that the output power of the panel is greatly affected by environmental factors such as light, and the panels far apart in the photovoltaic power generation system will have different output powers due to different environmental factors such as light, so that the power fluctuation can reflect the running state of the panel, and the value of the predetermined radius is 5, reducing the interference of environmental factors such as light.

[0036] Arranging the power fluctuation of each panel according to the coordinate position of the panel can obtain a power fluctuation map of the photovoltaic power generation system, wherein the power fluctuation map comprises the power fluctuation of each panel in the photovoltaic power generation system at the current moment, please refer to Figure 2 , which is a schematic diagram of the power fluctuation map according to an embodiment of the present application.​​​​

[0037] S102, threshold segmentation is performed on the power fluctuation map to obtain abnormal panels at the current time, and the abnormal panels are clustered to obtain abnormal clusters at the current time.

[0038] In an embodiment, in an ideal state, the power fluctuation map should be a uniformly distributed image with values close to 0 at each position, and when any panel is abnormal, the corresponding position will show a highlighted value. Therefore, threshold segmentation can well locate the abnormal panels at the current time.

[0039] Specifically, the threshold segmentation of the power fluctuation map includes determining a segmentation threshold of the power fluctuation map by using the maximum inter-class variance method, and marking panels with power fluctuation greater than the segmentation threshold as abnormal panels.

[0040] After obtaining the abnormal panels, only the abnormal panels are clustered to automatically divide the abnormal panels into multiple abnormal clusters, and each abnormal cluster includes at least one abnormal panel. Each abnormal cluster represents a continuous abnormal area. Specifically, the clustering of the abnormal panels to obtain the abnormal clusters at the current time includes initializing a neighborhood radius and a minimum sample number, and clustering the abnormal panels by using a DBSCAN algorithm to obtain multiple abnormal clusters.

[0041] The DBSCAN algorithm is a density-based clustering algorithm that can adaptively determine the number of abnormal clusters according to the distribution of the abnormal panels. The neighborhood radius is 1.5, and the minimum sample number is 3.

[0042] In this way, the distribution of each abnormal panel at the current time is obtained, and each abnormal cluster corresponds to a continuous abnormal area in the photovoltaic power generation system.

[0043] S103, in a plurality of consecutive historical time periods, a union set of any abnormal cluster is obtained, and the similarity of abnormal sequences between any panel and other panels in the union set is calculated, the similarity of the abnormal sequences is used as a weight to calculate the weighted correlation of output power between any panel and other panels, and the spatial correlation of each abnormal panel at the current time is obtained.

[0044] In an embodiment, each abnormal cluster covers all abnormal panels in the photovoltaic power generation system, and some abnormal panels are caused by environmental changes, such as shadows of clouds, shading of trees or animals, etc.; and some abnormal panels are caused by real faults. Therefore, in order to further distinguish real faults from pseudo faults caused by environmental changes, the abnormal clusters need to be further analyzed in time sequence.

[0045] The union set of any abnormal cluster includes: obtaining the union set of any abnormal cluster in the historical time period and the current time. The historical anomaly cluster with the largest intersection among the anomaly clusters is selected as the target cluster. Obtain the union of the target cluster and the abnormal cluster; then re-convert the historical time... The historical anomaly cluster with the largest intersection with the union of the given set is taken as the target cluster. The intersection is updated until the union of the abnormal clusters over multiple consecutive historical time periods is obtained.

[0046] It should be noted that since environmental factors are constantly changing, they can cause changes in the position and shape of anomalous clusters over time. By taking the union of any anomalous cluster over multiple consecutive historical moments, it can be ensured that all solar panels within the union are affected by the same environmental factors. For example, the shadow of a cloud moves over time, and the anomalous clusters caused by the shadow of the cloud will also move in the power fluctuation diagram. By taking the union of the anomalous clusters over multiple consecutive historical moments, all solar panels affected by the shadow of the cloud can be obtained.

[0047] Since all solar panels within a union are affected by the same environmental factors, if the faults of the solar panels within the union are pseudo-faults caused by environmental changes, the output power of each solar panel should show a strong correlation.

[0048] Obtain the anomaly sequences of each solar panel within the union set. An anomaly sequence for a single solar panel includes category labels indicating whether it belongs to an abnormal panel category across multiple consecutive historical time points. The anomaly sequence describes when the solar panel was affected by environmental factors and characterizes the temporal changes in the impact of environmental factors on the solar panel. For example, for the solar panels within the union set... In other words, if multiple consecutive historical moments are historical moment 1 to historical moment 5, and if the solar panel This is an abnormal solar panel. The category label is 1, otherwise, if the solar panel This is not an abnormal solar panel. The category label is 0; thus, the solar panel can be obtained. The abnormal sequence is This indicates the solar panels at historical moments 1 and 5. These are anomalous solar panels, occurring between historical time 2 and historical time 4. This is not an abnormal solar panel.

[0049] The similarity of the anomalous sequences is negatively correlated with their Hamming distance. (Solar panel) With other solar panels in the union Similarity between anomalous sequences Satisfying the relation:

[0050] ;

[0051] wherein, is the battery panel is the battery panel is the Hamming distance between abnormal sequences between the battery panel is the number of historical time instants, is an integer greater than 1. The similarity is greater, the more consistent the abnormality of the battery panel is with the abnormality of other battery panels in the set. is the same, i.e., the more consistent the time instants at which the two battery panels are marked as abnormal, the more similar the time patterns affected by the same environmental factor.

[0052] is the battery panel is the weighted correlation of output power between the battery panel satisfies the relationship:

[0053] ;

[0054] wherein, is the battery panel is the battery panel is the similarity of abnormal sequences between the battery panel is the battery panel is the sum of similarity of abnormal sequences between the battery panel is the battery panel is the battery panel is the Pearson correlation coefficient of output power between the battery panel is 0, the weighted correlation is denoted as 0.

[0055] It can be understood that by using as the weight, the weighted correlation pays more attention to the correlation between battery panels whose time patterns remain consistent, eliminates the reduction of output power correlation caused by the difference of abnormal sequences within the set, and accurately measures the correlation of output power between the battery panel and other battery panels within the set. The greater the weighted correlation of the battery panel is, the stronger the correlation of output power between the battery panel and each battery panel within the set is, indicating that the abnormality of the battery panel is consistent with other battery panels within the set, and it is more likely to be a false fault caused by environmental factors.

[0056] S104, according to the spatial correlation of the abnormal battery panel and the change rate of power fluctuation in the continuous multiple historical time instants, the abnormal battery panel at the current time is divided into a real fault and a false fault caused by environmental changes, and an abnormality warning is issued.

[0057] In one embodiment, if the abnormal panel is a real fault, the change rate of the power fluctuation of the abnormal panel should be maintained at a large level in a plurality of continuous historical time points, and if the abnormal panel is a pseudo fault caused by environmental changes, the change rate of the power fluctuation of the abnormal panel will fluctuate due to changes in environmental factors. Therefore, the type of fault of the abnormal panel can also be determined by the change rate of the power fluctuation of the abnormal panel. Specifically, the abnormal panel The change rate of the power fluctuation is the variance of the power fluctuation of the abnormal panel in a plurality of continuous historical time points.

[0058] According to the spatial correlation and the change rate of the power fluctuation of the abnormal panel, the abnormal panel at the current time point is divided into a real fault and a pseudo fault caused by environmental changes, including: in response to the spatial correlation being less than a correlation threshold and the change rate of the power fluctuation being less than a change rate threshold, determining that the abnormal panel is a real fault, and issuing an abnormal warning; otherwise, determining that the abnormal panel is a pseudo fault caused by environmental changes. Please refer to Figure 3 The identification result of the real fault and the pseudo fault caused by environmental changes according to the embodiments of the present application.

[0059] By jointly analyzing the change rate of the power fluctuation and the spatial correlation, different types of faults can be effectively distinguished, and an abnormal warning can be issued when the abnormal panel is a real fault, reducing the false positives of the abnormal warning caused by environmental interference, and improving the accuracy of the abnormal warning.

[0060] According to a second aspect of the present application, the present application also provides a power generation abnormal warning system for a photovoltaic power generation system. Figure 4 is a structural block diagram of the power generation abnormal warning system for a photovoltaic power generation system according to the embodiments of the present application. As Figure 4 shown, the system 50 includes a processor and a memory, and the memory stores computer program instructions, which, when executed by the processor, implement the power generation abnormal warning method for a photovoltaic power generation system according to the first aspect of the present application. The system also includes a communication bus, a communication interface and other components familiar to those skilled in the art, the settings and functions of which are known in the art, and therefore will not be described here.

[0061] It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of protection of the present application.​

Claims

1. A power generation anomaly early warning method for a photovoltaic power generation system, characterized by, The early warning method comprises: calculating the absolute value of the difference between the output power of any battery panel and the average value of the output power of each adjacent battery panel to obtain the power fluctuation of the battery panel; arranging the power fluctuation of each battery panel according to the coordinate position of the battery panel to obtain a power fluctuation map of the photovoltaic power generation system composed of the power fluctuation of each battery panel; The power fluctuation map is threshold segmented to obtain abnormal battery panels at the current moment, and the abnormal battery panels are clustered to obtain abnormal clusters at the current moment; At continuous multiple historical moments, the union of any abnormal cluster is obtained, and the similarity of abnormal sequences between any battery panel and other battery panels in the union is calculated in the union, the similarity of the abnormal sequences is taken as a weight to calculate the weighted correlation of the output power between any battery panel and other battery panels, and the spatial correlation of each abnormal battery panel at the current moment is obtained, wherein the abnormal sequence includes the category label of whether the battery panel belongs to an abnormal battery panel at continuous multiple historical moments; According to the spatial correlation of the abnormal battery panel and the change rate of the power fluctuation at continuous multiple historical moments, the abnormal battery panel at the current moment is divided into a real fault and a pseudo fault caused by environmental changes, and an abnormal early warning is issued; The change rate of the power fluctuation of the abnormal battery panel is the variance of the power fluctuation of the abnormal battery panel at continuous multiple historical moments.

2. The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized by, Obtaining the union of any anomaly clusters includes: combining historical moments In the present moment The historical anomaly cluster with the largest intersection among the anomaly clusters is selected as the target cluster. Obtain the union of the target cluster and the abnormal cluster; then re-convert the historical time... The historical anomaly cluster with the largest intersection with the union of the given set is taken as the target cluster. The intersection is updated until the union of the abnormal clusters over multiple consecutive historical time periods is obtained. 3.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, The similarity of the abnormal sequence is negatively correlated with the Hamming distance of the abnormal sequence. 4.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, The adjacent battery panel is a battery panel in a circular region with the battery panel as the center and a predetermined radius, except the battery panel. 5.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, Threshold segmentation of the power fluctuation map comprises: determining the segmentation threshold of the power fluctuation map by using the maximum inter-class variance method, and marking the battery panel with a power fluctuation greater than the segmentation threshold as an abnormal battery panel. 6.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, Battery panel Weighted correlation of output power between other battery panels Satisfies the relationship: ； wherein, is the similarity of the abnormal sequence between the battery panel and the other battery panel, is the similarity of the abnormal sequence between the battery panel and each of the other battery panels, is the number of the other battery panels in the union set, is the sum of the similarity of the abnormal sequence between the battery panel and each of the other battery panels, is the Pearson correlation coefficient of the output power between the battery panel and the other battery panel. 7.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, Clustering the abnormal battery panels to obtain abnormal clusters at the current moment comprises: initializing the neighborhood radius and the minimum number of samples, and clustering the abnormal battery panels by using the DBSCAN algorithm to obtain multiple abnormal clusters. 8.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, According to the spatial correlation of the abnormal battery panel and the change rate of the power fluctuation, the abnormal battery panel at the current moment is divided into a real fault and a pseudo fault caused by environmental changes, which comprises: in response to the spatial correlation being less than a correlation threshold and the change rate of the power fluctuation being less than a change rate threshold, determining that the abnormal battery panel is a real fault and issuing an abnormal early warning; otherwise, determining that the abnormal battery panel is a pseudo fault caused by environmental changes. 9.The power generation anomaly early warning method for a photovoltaic power generation system according to claim 1, characterized in that, When the battery panel belongs to an abnormal battery panel, the category label of the battery panel is 1, otherwise, the category label of the battery panel is 0.

10. A power generation abnormality early warning system for a photovoltaic power generation system, characterized by, The device comprises a processor and a memory, and the memory stores computer program instructions which, when executed by the processor, implement the power generation abnormal early warning method for a photovoltaic power generation system according to any one of claims 1 to 9.

Images