A Fault Diagnosis Method and System for Photovoltaic Inverter

Abstract

The present invention relates to the field of fault diagnosis technology, specifically to a fault diagnosis method and system for a photovoltaic inverter, comprising the following steps: based on an operating status data set, analyzing a variety of inverter operating data, identifying the volatility of a variety of operating data, extracting statistical features and frequency domain features of the data, adjusting the detection threshold of abnormal data, and generating abnormal fluctuation detection parameters. In the present invention, the abnormal detection threshold is dynamically adjusted by analyzing the operating data, a more sensitive response is made to the data, the accuracy and response speed of the fault warning are improved, the time window segmentation and signal feature extraction are used to timely discover abnormal fluctuations, detect equipment damage or system failures, and the accuracy of fault type identification is improved by combining the comparison and analysis of data features. By analyzing the component types associated with various fault types, the efficiency and accuracy of locating faulty components are improved, and the reliability and economic benefits of the photovoltaic inverter are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method and system for a photovoltaic inverter. Background Art

[0002] The field of fault diagnosis technology involves a variety of methods and tools to detect, analyze and identify faults or anomalies in systems, equipment and machines, determine the nature and location of the fault through a variety of diagnostic technologies, and perform effective maintenance and repair to reduce downtime and improve system reliability and efficiency. It includes electrical signal analysis, physical parameter monitoring, and system performance evaluation. It is applied to industrial automation, electronic manufacturing, automotive engineering and energy equipment management. It combines signal processing, pattern recognition and machine learning technologies to improve system reliability and reduce operational risks.

[0003] Among them, the fault diagnosis method of photovoltaic inverters focuses on the detection and analysis of electrical and performance faults that occur in photovoltaic inverter systems. It covers the use of diagnostic equipment to monitor the operating parameters of the inverter, including current, voltage and temperature, and identifies the fault point through abnormal fluctuations in multiple parameters. By real-time monitoring and analysis of inverter operating data, it helps operators quickly locate the source of the fault and take appropriate maintenance or replacement measures to ensure the stable operation and maximum power generation efficiency of the photovoltaic system, aiming to improve the safety and economy of photovoltaic power generation.

[0004] Under rapidly changing operating conditions, traditional photovoltaic inverter fault diagnosis methods have difficulty capturing subtle fault signals in real time or providing early fault warnings, and cannot meet the needs of efficient operation and maintenance. Fixed thresholds lead to insufficient recognition of new abnormal types and lack of adaptive capabilities. When a fault occurs, the faulty component cannot be located quickly and accurately, resulting in shutdown of the entire system, causing significant economic losses and safety risks in energy equipment management. There is a lack of ability to deeply analyze the impact of faults, and it is impossible to effectively evaluate the actual impact of faults on system performance, which affects decision-making in fault handling. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a fault diagnosis method and system for a photovoltaic inverter.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, a fault diagnosis method for a photovoltaic inverter, comprising the following steps: S1: Based on the operating status data set, analyze various operating data of the inverter, identify the volatility of various operating data, extract the statistical characteristics and frequency domain characteristics of the data, adjust the detection threshold of abnormal data, and generate abnormal fluctuation detection parameters; S2: Based on the abnormal fluctuation detection parameter, the current and voltage signals of the inverter are divided into time windows, the signal features of multiple time periods are extracted, the abnormal fluctuation data are identified by comparing with the abnormal detection threshold, and the abnormal degree of data at multiple time points is calculated to generate an abnormal detection result; S3: using the anomaly detection result, by analyzing the correlation between multiple data features and fault events, evaluating the feature importance of the abnormal data, extracting and analyzing key features, identifying the fault type by matching the abnormal features with known fault features, and generating a fault type classification result; S4: According to the fault type classification result, analyze the type of faulty component, and perform spectrum analysis on the current and voltage signals of the target type component in the inverter. By comparing with the baseline spectrum, identify the faulty component and mark the location information, and generate the faulty component positioning result.

[0007] As a further solution of the present invention, the steps of acquiring the statistical characteristics and frequency domain characteristics of the data are specifically as follows: S111: Based on the running status data set, through the formula: ; Calculate the Euclidean distance between data points to get the volatility indicator; in, represents the volatility index, represents the total number of data points, Indicates data points, Indicates data points, is the index of the data point, is the weight coefficient for adjusting the mean deviation, is the global average of the data set; S112: Based on the volatility index, extract statistical features of various operating data, including mean, median, variance and standard deviation, quantify the central tendency and dispersion of the data, and generate preliminary feature extraction results; S113: Based on the preliminary feature extraction results, use Fourier transform to analyze the frequency domain characteristics and periodic changes of the data, identify key frequency components in the data, and obtain statistical characteristics and frequency domain characteristics of the data.

[0008] As a further solution of the present invention, the step of obtaining the abnormal fluctuation detection parameter is specifically as follows: S121: Based on the statistical characteristics and frequency domain characteristics of the data, identifying the volatility baselines of the various operating data, and generating volatility baseline data; S122: Based on the volatility baseline data, by formula: ; Calculate the adjusted abnormal fluctuation detection threshold to obtain abnormal fluctuation detection parameters; in, is the adjusted abnormal fluctuation detection threshold, is the original threshold value based on the volatility baseline, is the adjustment factor, is the number of historical data points, For the Historical fluctuation data points, is the average value of historical fluctuation data, is the index.

[0009] As a further solution of the present invention, the step of acquiring the abnormal fluctuation data is specifically as follows: S211: Based on the abnormal fluctuation detection parameter, time window segmentation is performed on the current and voltage signals of the inverter, and continuous signal data is divided into multiple sequences to generate a time window data set; S212: extracting signal features in multiple time windows based on the time window data set, analyzing signal fluctuations in the multiple windows, and obtaining a signal feature data set; S213: Based on the signal feature data set, by formula: ; The weighted standardized deviation in each time window is calculated and compared with the preset anomaly detection threshold to obtain the abnormal fluctuation data set; is the weighted standardized deviation, is the value of a single data point in the time window, is the average value of the data points in the time window, To allocate to The weight of the data point, is the total number of data points in the time window, Represents the index of the data point within the target time window.

[0010] As a further solution of the present invention, the step of obtaining the abnormality detection result is specifically: S221: extracting various operating data in the abnormal fluctuation time window based on the abnormal fluctuation data, including current and voltage readings, and generating a data feature set; S222: Based on the data feature set, by formula: ; Calculate the data anomaly degree score of each time window to obtain an anomaly degree score set; in, is the abnormality score of the time window, is the total number of data points in the time window, is the index of the data point in the time window, The time window The value of the data point, is the average value of the time window data, is the standard deviation of the data points in the time window, is the preset anomaly detection threshold; S223: Based on the abnormality degree score set, record the time, abnormal value type, and abnormality degree information of the target abnormal data point to generate an abnormality detection result.

[0011] As a further solution of the present invention, the step of obtaining the fault type classification result is specifically: S311: Based on the abnormality detection result, by analyzing the correlation between multiple data features and fault events, performing feature importance evaluation on multiple data, and generating feature importance analysis results; S312: Based on the feature importance analysis result, the formula is: ; Calculate the similarity between the key features and the known fault features to obtain the feature matching score; in, represents the cosine similarity, is extracted from the abnormal data Key features, This is the first Features, is the total number of key features for comparison, The index representing the feature; S313: Based on the result of the feature matching analysis, the matching scores of multiple fault types are compared to identify the fault type and generate a fault type classification result.

[0012] As a further solution of the present invention, the step of obtaining the fault component positioning result is specifically: S411: Based on the fault type classification result, analyzing the fault component type corresponding to the target fault type, and generating a component type analysis result; S412: According to the component type analysis result, use the formula: ; Calculating frequency spectrum data of power signals of multiple components; in, Indicates the frequency The spectrum amplitude at It is the first The signal value of the sampling point, is the total number of samples in the signal, is the current sampling point index, is the specific frequency analyzed, is the base of natural logarithms, is an imaginary unit, is the ratio of pi; S413: According to the spectrum data of the power signals of the multiple components, by comparing with the baseline spectrum, the faulty component is identified, and the position information of the faulty component is marked to generate a faulty component positioning result.

[0013] A photovoltaic inverter fault diagnosis system, the photovoltaic inverter fault diagnosis system is used to perform the photovoltaic inverter fault diagnosis method, the system comprises: The data analysis module performs volatility analysis on various operating data based on the operating status data set, extracts the statistical and frequency domain features of the data, adjusts the detection threshold of abnormal data, and generates abnormal fluctuation detection parameters; The anomaly identification module extracts signal features based on the abnormal fluctuation detection parameters by time window segmentation, compares the signal features with the anomaly detection threshold, identifies the data points of abnormal fluctuations, and calculates the degree of data anomalies at multiple time points to generate anomaly detection results. The fault classification module performs importance assessment of fault data features and key feature extraction based on the abnormal detection results, identifies the target fault type by matching the abnormal feature data with known fault feature data, and generates a fault type classification result; The component locating module analyzes component types associated with multiple fault types based on the fault type classification result, performs spectrum analysis on current and voltage signals of target type components in the inverter, identifies faulty components and marks location information, and generates faulty component locating results.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the abnormal detection threshold is dynamically adjusted by analyzing the operating data, and a more sensitive response is made to the data, thereby improving the accuracy and response speed of fault warning. Time window segmentation and signal feature extraction are used to timely discover abnormal fluctuations, detect equipment damage or system failures, and combine the comparison and analysis of data features to improve the accuracy of fault type identification. By analyzing the component types associated with various fault types, the efficiency and accuracy of locating faulty components are improved, thereby enhancing the reliability and economic benefits of photovoltaic inverters. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2It is a flow chart of obtaining statistical characteristics and frequency domain characteristics of data of the present invention; Figure 3 This is a flow chart of obtaining abnormal fluctuation detection parameters of the present invention; Figure 4 The flowchart of obtaining abnormal fluctuation data of the present invention; Figure 5 The flowchart of obtaining abnormal detection results of the present invention; Figure 6 A flow chart of obtaining fault type classification results of the present invention; Figure 7 This is a flow chart of obtaining the fault component location result of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying 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 limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 The present invention provides a technical solution, a fault diagnosis method for a photovoltaic inverter, comprising the following steps: S1: Based on the operating status data set, analyze various operating data of the inverter, identify the volatility of various operating data, extract the statistical characteristics and frequency domain characteristics of the data, adjust the detection threshold of abnormal data, and generate abnormal fluctuation detection parameters; S2: Based on the abnormal fluctuation detection parameters, the current and voltage signals of the inverter are divided into time windows, the signal features of multiple time periods are extracted, and the abnormal fluctuation data is identified by comparing with the abnormal detection threshold. The degree of data abnormality at multiple time points is calculated to generate an abnormal detection result; S3: Using the anomaly detection results, by analyzing the correlation between multiple data features and fault events, the feature importance of the abnormal data is evaluated, key features are extracted and analyzed, and the fault type is identified by matching the abnormal features with known fault features to generate a fault type classification result; S4: According to the fault type classification result, analyze the faulty component type, and perform spectrum analysis on the current and voltage signals of the target type components in the inverter. By comparing with the baseline spectrum, identify the faulty component and mark the location information, and generate the faulty component location result.

[0019] The abnormal fluctuation detection parameters specifically include the statistical characteristics of the data, the frequency domain characteristics of the data, and the detection threshold parameters. The abnormal detection results include abnormal fluctuation data, abnormality degree score, and time window feature set. The fault type classification results specifically refer to key fault characteristics, fault matching degree, and fault type identification. The fault component location results include fault component type, spectrum comparison results, and location information mark.

[0020] See also Figure 2 , the steps for obtaining the statistical characteristics and frequency domain characteristics of the data are as follows: S111: Based on the running status data set, through the formula: ; Calculate the Euclidean distance between data points to get the volatility indicator; in, represents the volatility index, represents the total number of data points, Indicates data points, Indicates data points, is the index of the data point, is the weight coefficient for adjusting the mean deviation, is the global average of the data set; formula: ; Parameter meaning and acquisition method: : is the total number of data points, obtained by directly reading the number of data points from the inverter monitoring system; :For The voltage or current value of each data point is collected at a specific time point by the inverter monitoring system; :For The voltage or current value of each data point is monitored by the inverter monitoring system. The next time point after the acquisition is obtained; : is the weight coefficient, which is used to adjust the impact caused by mean deviation; : is the global average value of the data set, obtained by calculating the average value of all data points collected by the inverter monitoring system; Calculation example: set up , , , , substitute the set value into the calculation: The first pair of data points and : ; ; ; ; The second pair of data points and : ; ; ; ; The third pair of data points and : ; ; ; ; The fourth pair of data points and : ; ; ; ; calculate : ; ; Calculation results It shows the average volatility between each pair of adjacent data points. The value reflects the instability of voltage or current data. The value shows the stability of inverter operation. The calculation process is used to analyze the volatility of various operating data under normal operating conditions and determine the abnormal data detection threshold.

[0021] S112: Based on volatility indicators, extract statistical features of various operating data, including mean, median, variance and standard deviation, quantify the central tendency and dispersion of the data, and generate preliminary feature extraction results; In sub-step S112, the core statistical indicators of the inverter operation data are extracted and analyzed. The process includes calculating the mean, median, variance and standard deviation of the operation data. Each indicator quantifies the central trend and dispersion of the data. The mean provides the average level of the data, the median gives the center point of the data distribution, and the variance and standard deviation reveal the fluctuation range and consistency of the data. The statistical indicators are synthesized into a set, which provides the necessary basic information for the subsequent frequency domain feature analysis. By performing statistical calculations on the data of each operation cycle, the performance of the equipment under different operating conditions is evaluated, providing data support for equipment maintenance and optimization.

[0022] S113: Based on the preliminary feature extraction results, use Fourier transform to analyze the frequency domain characteristics and periodic changes of the data, identify the key frequency components in the data, and obtain the statistical characteristics and frequency domain characteristics of the data; In sub-step S113, Fourier transform is used to perform frequency domain feature analysis on the data. The process covers frequency component identification and analysis of the periodic changes and spectral density of the data. Fourier transform converts time series data into frequency domain representation, revealing the main frequency components in the data, which is crucial for understanding the root causes of data fluctuations. During the analysis process, the power spectral density of the data is calculated to identify the key frequency points that affect the performance of the equipment. The target frequency domain features and statistical features together constitute a comprehensive data feature description, providing in-depth insights into the operating status of the equipment, which helps to accurately diagnose possible operating anomalies or foresee potential equipment failures.

[0023] See also Figure 3 , the specific steps for obtaining abnormal fluctuation detection parameters are: S121: Based on the statistical characteristics and frequency domain characteristics of the data, identify the volatility baselines of various operating data and generate volatility baseline data; In sub-step S121, detailed identification and data generation of the volatility baseline are performed. A comprehensive volatility baseline is established by integrating statistical indicators such as the mean, median, variance and standard deviation in the data set with the frequency components obtained by Fourier transform analysis. The baseline reflects the data fluctuation pattern of the inverter under normal operating conditions and provides a reference standard for anomaly detection and health monitoring. In addition, a variety of algorithms, such as moving average and exponential smoothing, are combined to balance the changes in short-term and long-term data to ensure the stability and applicability of the baseline. By comparing real-time data with this baseline, data points that deviate from the normal range are identified, and potential equipment problems or performance degradation are discovered in a timely manner.

[0024] S122: Based on the volatility baseline data, the formula is: ; Calculate the adjusted abnormal fluctuation detection threshold to obtain abnormal fluctuation detection parameters; in, is the adjusted abnormal fluctuation detection threshold, is the original threshold based on the volatility baseline, is the adjustment factor, is the number of historical data points, For the Historical fluctuation data points, is the average value of historical fluctuation data, is the index; formula: ; Parameter meaning and acquisition method: : Benchmark threshold, obtained from the analysis of historical fluctuation data; : Adjusted abnormal fluctuation detection threshold; : Adjustment coefficient, set through sensitivity analysis of the system; : No. Historical fluctuation data points are obtained from the database of the monitoring system; : The average value of historical data, obtained by calculation; : The total number of historical data points; Calculation example: set up , , , The values ​​are [2,3,5,6,9,11,3,2,0,8], ; Calculation steps: ; ; ; Calculation results The adjusted abnormal fluctuation detection threshold is shown. The calculation process is used to optimize the inverter fault detection system to ensure that the alarm is triggered when the fluctuation increases significantly.

[0025] See also Figure 4 ,The specific steps for obtaining abnormal fluctuation data are: S211: Based on the abnormal fluctuation detection parameter, the current and voltage signals of the inverter are divided into time windows, and the continuous signal data is divided into multiple sequences to generate a time window data set; In sub-step S211, the process involves accurately segmenting the continuous signal data into multiple independent time series so that the data characteristics of each series can be analyzed separately. To achieve the segmentation, the sliding window technology is used, in which the window size and step size are set according to the specific operating frequency of the inverter and the expected monitoring requirements. Through this method, a data set containing multiple time windows is generated, each window contains current and voltage data over a period of time. This data set will be used to further analyze the signal fluctuations and abnormal patterns in each time window, and provide real-time monitoring of the equipment status and abnormal warning.

[0026] S212: extracting signal features in multiple time windows based on the time window data set, analyzing signal fluctuations in the multiple windows, and obtaining a signal feature data set; In sub-step S212, the current and voltage signal characteristics in each time window are extracted and analyzed. The feature extraction methods used include peak analysis, mean calculation, and waveform integration, which reveal the signal fluctuations and potential irregularities in each time window. After summarizing the analysis results of each time window, a signal feature data set is obtained. The data set records the performance of the signal in each segmented window, including the fluctuation amplitude, frequency component and its changing trend. The target information is crucial for judging the operating status of the equipment and early fault diagnosis, helping the technical team to effectively intervene in the early stages of the problem.

[0027] S213: Based on the signal feature data set, through the formula: ; The weighted standardized deviation in each time window is calculated and compared with the preset anomaly detection threshold to obtain the abnormal fluctuation data set; is the weighted standardized deviation, is the value of a single data point in the time window, is the average value of the data points in the time window, To allocate to The weight of the data point, is the total number of data points in the time window, Represents the index of the data point in the target time window; formula: ; Parameter meaning and acquisition method: : represents the value of a single data point in the time window, which is obtained by real-time monitoring of the current and voltage signals of the inverter; : represents the average value of the data points in the time window, which is obtained by calculating the average value of all data points in the time window; :Represents the allocation to The weight of each data point; : represents the total number of data points in the time window, which is obtained by counting the number of data points in each time window; : Represents the preset anomaly detection threshold, which is set based on historical data analysis or equipment failure records; : represents the calculated standardized deviation value; Calculation example: set up , , , , ; Specific calculation process of the formula: ; ; The calculation result is 0.229, indicating that the fluctuation of the data points in the target time window does not exceed the set anomaly detection threshold. The result reflects that under the currently set threshold and data conditions, the current or voltage data in the time window exhibits normal fluctuations. The calculation process is used to detect and identify abnormal fluctuations in current or voltage data.

[0028] See also Figure 5 , the specific steps for obtaining anomaly detection results are: S221: extracting various operating data in the abnormal fluctuation time window based on the abnormal fluctuation data, including current and voltage readings, and generating a data feature set; In sub-step S221, multi-dimensional data feature extraction technology is used, including signal peak analysis, minimum, maximum and mean calculation, to calculate the skewness and kurtosis of the current and voltage readings, identify the sharpness and skew direction of the data distribution, and summarize all target data features into a data feature set, which will serve as the basis for subsequent diagnosis and analysis. The key to this process is to use specific statistical tools and mathematical models to describe and quantify the current and voltage data within the time window in detail, ensuring accurate capture of data and accurate identification of abnormal features.

[0029] S222: Based on the data feature set, through the formula: ; Calculate the data anomaly degree score of each time window to obtain an anomaly degree score set; in, is the abnormality score of the time window, is the total number of data points in the time window, is the index of the data point in the time window, The time window The value of the data point, is the average value of the time window data, is the standard deviation of the data points in the time window, is the preset anomaly detection threshold; formula: ; Parameter meaning and acquisition method: : is the total number of data points in the time window, obtained by counting the data points in the time window; : is the index of the data point in the time window; : is the first The value of each data point is obtained through the data acquisition system of the monitoring equipment; : is the average value of the time window data, obtained by calculating the average value of all data points in the time window; : is the standard deviation of the data points in the time window, calculated by statistical methods; : is the preset anomaly detection threshold; Calculation example: set up , , , ; ; Substitute the set value into the calculation: ; ; ; ; The calculated result of 0.016 reflects the overall abnormality of the data in the time window, which is relatively low. The value indicates that the fluctuation of the data points in the time window is within the acceptable range and is consistent with the expected fluctuation under normal operating conditions. The calculation process is used to evaluate and quantify the degree of data abnormality within a specific time window to help identify potential equipment or system problems.

[0030] S223: Based on the abnormality degree score set, record the time, abnormal value type, and abnormality degree information of the target abnormal data point, and generate an abnormality detection result; In sub-step S223, the specific timestamp of each data point identified as abnormal is recorded, the type of abnormal value, such as current or voltage outside the normal range, and the degree of abnormality of the data point. The recording work uses a database management system to store and manage target key information to facilitate historical data comparison and trend analysis. All collected information is compiled into an abnormality detection result. The report lists the details of all abnormal events in detail, including the time when the abnormality occurred, the components involved, the specific characteristics of the abnormality and the possible cause analysis. The report is crucial for the technical team to track equipment status, analyze the cause of failure and formulate maintenance plans.

[0031] See also Figure 6 , the specific steps for obtaining the fault type classification results are: S311: Based on the anomaly detection result, by analyzing the correlation between various data features and fault events, feature importance evaluation is performed on various data to generate feature importance analysis results; In sub-step S311, feature selection methods in machine learning technology, such as the random forest algorithm, are used to evaluate the importance of each data feature in the fault diagnosis model, including calculating the contribution of each feature to the fault prediction model, including evaluating the importance through the feature's information gain and Gini impurity reduction. Through target technology, feature importance analysis results are generated, listing the importance level of each data feature in diagnosing different types of faults, helping the technical team identify the data features that are most critical to fault prediction, optimizing the subsequent fault diagnosis process, and improving the accuracy and efficiency of fault diagnosis.

[0032] S312: Based on the feature importance analysis results, the formula is: ; Calculate the similarity between the key features and the known fault features to obtain the feature matching score; in, represents the cosine similarity, is extracted from the abnormal data Key features, This is the first Features, is the total number of key features for comparison, The index representing the feature; formula: ; Parameter meaning and acquisition method: : Cosine similarity, quantifies the similarity between feature vectors; : Key features extracted from abnormal data; : Features associated with known faults; : The total number of key features; Calculation example: set up , , ; Specific calculation process of the formula: ; ; ; ; ; Calculation results It shows that there is a very high similarity between the key features of abnormal data and known fault features, and the calculation process is used to provide a data basis for fault diagnosis and classification.

[0033] S313: Based on the result of feature matching analysis, the matching scores of multiple fault types are compared to identify the fault type and generate a fault type classification result; In sub-step S313, the possibility of each fault type is scored by using statistical analysis methods and classification algorithms in machine learning, including support vector machines. The step involves calculating the similarity between each fault feature vector and the known fault type feature template. The similarity calculation methods include cosine similarity and Euclidean distance. After the matching score of each fault type is obtained, it is sorted according to the score to identify the fault type. The generated fault type classification result provides a probability assessment of each fault, which is an important basis for maintenance personnel to diagnose and handle faults. The process ensures the scientificity and systematicness of the diagnostic results and improves the response speed and accuracy of fault handling.

[0034] See also Figure 7 , the specific steps for obtaining the fault component location result are: S411: Analyze the fault component type corresponding to the target fault type based on the fault type classification result, and generate a component type analysis result; In sub-step S411, a decision tree classification model is used. The model infers the component type most likely to be affected based on the specific characteristics of the fault type. During the analysis process, the correlation between each fault type and the failure mode of a specific component is recorded, such as the relationship between capacitor failure and voltage drop, and the correlation between transformer failure and current anomaly. The target data is obtained through historical fault records and model learning, which ensures the accuracy and reliability of the analysis results. The component type analysis results are generated, which lists the components that may be affected by each fault type and evaluates the probability of failure of each component, providing important decision support for subsequent fault diagnosis and maintenance work.

[0035] S412: Analyze the results based on the component type, using the formula: ; Calculating frequency spectrum data of power signals of multiple components; in, Indicates the frequency The spectrum amplitude at It is the first The signal value of the sampling point, is the total number of samples in the signal, is the current sampling point index, is the specific frequency analyzed, is the base of natural logarithms, is an imaginary unit, is the ratio of pi; formula: ; Parameter meaning and acquisition method: : Total number of sampling points; : sampling point index; : At the sampling point The signal value of : target frequency; : the base of natural logarithms; : imaginary unit; : twice pi, used to convert frequency to angular frequency; Calculation example: set up , , for all , , substitute the set value into the calculation: ; ; , considering all are all 1, and add up: ; ; ; ; Calculation results It means that the signal strength under the set conditions is 0.2472. The result helps determine whether there is an abnormal signal in the target component, identify the specific faulty component, and provide key information for maintenance and repair.

[0036] S413: identifying the faulty component based on the spectrum data of the power signals of the multiple components by comparing with the baseline spectrum, marking the location information of the faulty component, and generating a faulty component location result; In sub-step S413, the fast Fourier transform technology is used to convert the time domain signal into a frequency domain signal. During the analysis process, the actual spectrum data is compared in detail with the pre-defined baseline spectrum. The baseline spectrum is a standard spectrum obtained from components in normal operating status. Through the difference analysis of the spectrum, the specific frequency components in the abnormal signal are identified. The target components indicate possible faulty components. The identified faulty components and their exact locations are marked in the system diagram, and the faulty component positioning results are generated. The results provide the location information of the specific faulty components, describe the nature and possible causes of the fault, and facilitate the maintenance team to respond to the fault quickly and accurately.

[0037] A photovoltaic inverter fault diagnosis system, the photovoltaic inverter fault diagnosis system is used to execute the photovoltaic inverter fault diagnosis method, the system comprises: The data analysis module performs volatility analysis on various operating data based on the operating status data set, extracts the statistical and frequency domain features of the data, adjusts the detection threshold of abnormal data, and generates abnormal fluctuation detection parameters; The anomaly recognition module extracts signal features based on the abnormal fluctuation detection parameters through time window segmentation, compares them with the anomaly detection threshold, identifies data points with abnormal fluctuations, calculates the degree of data anomalies at multiple time points, and generates anomaly detection results; The fault classification module evaluates the importance of fault data features and extracts key features based on the anomaly detection results. It identifies the target fault type by matching the anomaly feature data with the known fault feature data and generates a fault type classification result. The component location module analyzes the component types associated with multiple fault types based on the fault type classification results, performs spectrum analysis on the current and voltage signals of the target type components in the inverter, identifies the faulty components and marks the location information, and generates the faulty component location results.

[0038] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A fault diagnosis method for a photovoltaic inverter, characterized in that: The following steps are involved: S1: Based on the operating status data set, analyze various operating data of the inverter, identify the volatility of various operating data, extract the statistical characteristics and frequency domain characteristics of the data, adjust the detection threshold of abnormal data, and generate abnormal fluctuation detection parameters; S2: Based on the abnormal fluctuation detection parameter, the current and voltage signals of the inverter are divided into time windows, the signal features of multiple time periods are extracted, the abnormal fluctuation data are identified by comparing with the abnormal detection threshold, and the abnormal degree of data at multiple time points is calculated to generate an abnormal detection result; S3: using the anomaly detection result, by analyzing the correlation between multiple data features and fault events, evaluating the feature importance of the abnormal data, extracting and analyzing key features, identifying the fault type by matching the abnormal features with known fault features, and generating a fault type classification result; S4: According to the fault type classification result, analyze the type of faulty component, and perform spectrum analysis on the current and voltage signals of the target type component in the inverter. By comparing with the baseline spectrum, identify the faulty component and mark the location information, and generate the faulty component positioning result.

2. The fault diagnosis method for photovoltaic inverter according to claim 1, characterized in that: The steps for obtaining the statistical characteristics and frequency domain characteristics of the data are specifically as follows: S111: Based on the running status data set, through the formula: ； Calculate the Euclidean distance between data points to get the volatility indicator; in, represents the volatility index, represents the total number of data points, Indicates data points, Indicates data points, is the index of the data point, is the weight coefficient adjusted for mean deviation, is the global average of the data set; S112: Based on the volatility index, extract statistical features of various operating data, including mean, median, variance and standard deviation, quantify the central tendency and dispersion of the data, and generate preliminary feature extraction results; S113: Based on the preliminary feature extraction results, use Fourier transform to analyze the frequency domain characteristics and periodic changes of the data, identify key frequency components in the data, and obtain statistical characteristics and frequency domain characteristics of the data.

3. The fault diagnosis method for photovoltaic inverter according to claim 2, characterized in that: The steps for obtaining the abnormal fluctuation detection parameters are specifically as follows: S121: Based on the statistical characteristics and frequency domain characteristics of the data, identifying the volatility baselines of the various operating data, and generating volatility baseline data; S122: Based on the volatility baseline data, by formula: ； Calculate the adjusted abnormal fluctuation detection threshold to obtain abnormal fluctuation detection parameters; in, is the adjusted abnormal fluctuation detection threshold, is the original threshold value based on the volatility baseline, is the adjustment factor, is the number of historical data points, For the Historical fluctuation data points, is the average value of historical fluctuation data, is the index.

4. The fault diagnosis method for photovoltaic inverter according to claim 1, characterized in that: The steps for obtaining the abnormal fluctuation data are specifically as follows: S211: Based on the abnormal fluctuation detection parameter, time window segmentation is performed on the current and voltage signals of the inverter, and continuous signal data is divided into multiple sequences to generate a time window data set; S212: extracting signal features in multiple time windows based on the time window data set, analyzing signal fluctuations in the multiple windows, and obtaining a signal feature data set; S213: Based on the signal feature data set, by formula: ； The weighted standardized deviation in each time window is calculated and compared with the preset anomaly detection threshold to obtain the abnormal fluctuation data set; is the weighted standardized deviation, is the value of a single data point in the time window, is the average value of the data points in the time window, To allocate to The weight of the data point, is the total number of data points in the time window, Represents the index of the data point within the target time window.

5. The fault diagnosis method for photovoltaic inverter according to claim 4, characterized in that: The steps for obtaining the abnormal detection result are specifically as follows: S221: extracting various operating data in the abnormal fluctuation time window based on the abnormal fluctuation data, including current and voltage readings, and generating a data feature set; S222: Based on the data feature set, by formula: ； Calculate the data anomaly degree score of each time window to obtain an anomaly degree score set; in, is the abnormality score of the time window, is the total number of data points in the time window, is the index of the data point in the time window, The time window The value of the data point, is the average value of the time window data, is the standard deviation of the data points in the time window, is the preset anomaly detection threshold; S223: Based on the abnormality degree score set, record the time, abnormal value type, and abnormality degree information of the target abnormal data point to generate an abnormality detection result.

6. The fault diagnosis method for photovoltaic inverter according to claim 1, characterized in that: The steps for obtaining the fault type classification result are specifically as follows: S311: Based on the abnormality detection result, by analyzing the correlation between multiple data features and fault events, performing feature importance evaluation on multiple data, and generating feature importance analysis results; S312: Based on the feature importance analysis result, the formula is: ； Calculate the similarity between the key features and the known fault features to obtain the feature matching score; in, represents the cosine similarity, is extracted from the abnormal data Key features, This is the first Features, is the total number of key features for comparison, The index representing the feature; S313: Based on the result of the feature matching analysis, the matching scores of multiple fault types are compared to identify the fault type and generate a fault type classification result.

7. The fault diagnosis method for photovoltaic inverter according to claim 1, characterized in that: The steps for obtaining the fault component location result are specifically as follows: S411: Based on the fault type classification result, analyzing the fault component type corresponding to the target fault type, and generating a component type analysis result; S412: According to the component type analysis result, use the formula: ； Calculating frequency spectrum data of power signals of multiple components; in, Indicates the frequency The spectrum amplitude at It is the first The signal value of the sampling point, is the total number of samples in the signal, is the current sampling point index, is the specific frequency analyzed, is the base of natural logarithms, is an imaginary unit, is the ratio of pi; S413: According to the spectrum data of the power signals of the multiple components, by comparing with the baseline spectrum, the faulty component is identified, and the position information of the faulty component is marked to generate a faulty component positioning result.

8. A photovoltaic inverter fault diagnosis system, characterized in that: According to the photovoltaic inverter fault diagnosis method according to any one of claims 1 to 7, the system comprises: The data analysis module performs volatility analysis on various operating data based on the operating status data set, extracts the statistical and frequency domain features of the data, adjusts the detection threshold of abnormal data, and generates abnormal fluctuation detection parameters; The anomaly identification module extracts signal features based on the abnormal fluctuation detection parameters by time window segmentation, compares the signal features with the anomaly detection threshold, identifies the data points of abnormal fluctuations, and calculates the degree of data anomalies at multiple time points to generate anomaly detection results. The fault classification module performs importance assessment of fault data features and key feature extraction based on the abnormal detection results, identifies the target fault type by matching the abnormal feature data with known fault feature data, and generates a fault type classification result; The component locating module analyzes component types associated with multiple fault types based on the fault type classification result, performs spectrum analysis on current and voltage signals of target type components in the inverter, identifies faulty components and marks location information, and generates faulty component locating results.

Images