A photovoltaic fault detection system and diagnostic method based on an adaptive strategy

Abstract

This invention discloses a photovoltaic fault detection system and diagnostic method based on an adaptive strategy. The system includes a data acquisition and preprocessing module, a fault diagnosis and classification module, a fault location impact analysis module, and an adaptive control and maintenance decision-making module. Data is collected in real time through sensors such as voltage, current, and temperature. A Kalman filter algorithm is used for dynamic denoising and standardization. Support vector machines and K-means clustering algorithms are used to identify fault modes, and regression analysis is used to predict trends. A weighted shortest path algorithm and spatial distribution analysis are combined to accurately locate the fault source. Finally, based on real-time data, equipment parameters are adaptively adjusted to generate preventative maintenance strategies and automatically prioritize them for automated maintenance. This invention effectively improves fault detection accuracy, reduces operation and maintenance costs, has strong anti-interference capabilities and real-time adaptability, and is suitable for complex photovoltaic environments.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic system fault detection technology, and in particular to a photovoltaic fault detection system and diagnostic method based on an adaptive strategy. Background Technology

[0002] Photovoltaic power plant fault detection is a technology applied to the operation and management of photovoltaic systems. It is mainly used to identify common faults in photovoltaic arrays and grid-connected inverters, including cell module cracking, line aging, hot spot phenomena, and inverter overvoltage and overcurrent. This technology achieves fault location through infrared image detection, electrical signal detection, event correlation detection, and sensor detection. Among these methods, infrared detection has insufficient accuracy, and electrical signal detection requires system shutdown.

[0003] In existing technologies, fault detection methods for photovoltaic systems mainly rely on traditional sensor monitoring and manual inspection. While these methods can identify faults to some extent, they often have some problems, such as high data noise, low fault identification accuracy, and poor real-time performance. At the same time, existing fault diagnosis methods are mostly single-dimensional analyses, making it difficult to process data from different sensors simultaneously. Furthermore, they face significant challenges in accurately identifying and locating complex faults. Traditional fault detection methods lack adaptability, cannot flexibly adjust operating parameters according to real-time system status and environmental changes, and cannot effectively predict fault trends. Consequently, when photovoltaic systems malfunction, effective maintenance measures cannot be taken in a timely manner, thereby increasing operation and maintenance costs and affecting the stability and long-term operating efficiency of photovoltaic systems. Summary of the Invention

[0004] The purpose of this invention is to provide a photovoltaic fault detection system and diagnostic method based on an adaptive strategy.

[0005] The technical solution adopted in this invention is:

[0006] A photovoltaic fault detection system based on an adaptive strategy includes:

[0007] The data acquisition and preprocessing module is used to collect real-time data on voltage, current, temperature and radiation intensity of photovoltaic modules through a multi-sensor network, and to obtain preliminary abnormal data by dynamically denoising and standardizing the collected data through a Kalman filter algorithm.

[0008] The fault diagnosis and classification module is connected to the data acquisition and preprocessing module. It is used to extract features from the processed acquisition data and perform fault pattern recognition and classification on the extracted features based on support vector machine and K-means clustering algorithm. At the same time, it predicts fault trends and risks through regression algorithm.

[0009] The fault location and impact analysis module, connected to the fault diagnosis and classification module, is used to locate the fault location and its impact range through spatial distribution analysis and weighted shortest path algorithm; assess the impact of the fault on the overall performance of the photovoltaic system and provide repair strategies.

[0010] The adaptive maintenance decision module, connected to the fault location impact analysis module, is used to dynamically adjust equipment operating parameters based on fault diagnosis and location results to generate preventive maintenance strategies, and automatically prioritize maintenance based on fault severity for automated repair.

[0011] Furthermore, the photovoltaic data acquisition and preprocessing module includes a data acquisition sensor interface module, a data cleaning and denoising module, a data standardization and normalization module, and a real-time data monitoring and anomaly detection module. The data acquisition sensor interface module is responsible for acquiring real-time data from photovoltaic panels and inverter hardware to obtain raw sensor data. The data cleaning and denoising module removes noise and outliers from the sensor data using a Kalman filter adaptive filter to obtain denoised data. The data standardization and normalization module standardizes data from different acquisition devices to obtain data with a unified standard, ensuring that different data dimensions can be processed under the same standard. The real-time data monitoring and anomaly detection module uses anomaly detection algorithms to identify potential problems in the acquired data to obtain preliminary anomaly data.

[0012] Furthermore, the fault diagnosis and classification module includes a data feature extraction module, a fault mode recognition module, a classification and clustering module, and a fault prediction trend analysis module. The data feature extraction module extracts current and voltage curves, radiation changes, and temperature changes from the collected data. The fault mode recognition module uses SVM to train the collected data features and determine the fault type. The classification and clustering module classifies the fault types using the K-means algorithm. The fault prediction trend analysis module uses a regression algorithm to predict the future state of the photovoltaic system and monitor potential fault trends, providing preventative maintenance support for the system.

[0013] Furthermore, the fault location impact analysis module includes a fault source location module, a system status assessment module, a module / array level analysis module, and a redundancy analysis module. The fault source location module locates the fault source by analyzing the spatial distribution of sensor data and inverter feedback information. The system status assessment module generates impact assessment results by monitoring the status of equipment to assess the impact of the fault. The module / array level analysis module analyzes the status of the photovoltaic array or individual modules to refine the fault location. The redundancy analysis module analyzes the system's redundancy design and calculates the redundancy mechanism response when a fault occurs, and evaluates the performance of the redundancy mechanism when a fault occurs to help formulate a repair strategy.

[0014] Furthermore, the adaptive control maintenance decision module includes an adaptive control algorithm module, a preventive maintenance decision support module, an automatic repair mechanism module, and a maintenance priority ranking module. The adaptive control algorithm module adjusts the operating parameters of the corresponding equipment (inverters and other equipment) in the photovoltaic system according to the real-time status of the photovoltaic system. The preventive maintenance decision support module formulates maintenance plans in advance based on fault prediction and trend analysis results. The automatic repair mechanism module automatically repairs minor faults of specified categories to restore system operation through remote operation or adaptive strategies. The maintenance priority ranking module prioritizes maintenance tasks by assessing the degree of fault impact and recovery difficulty.

[0015] This invention also discloses a photovoltaic fault diagnosis method based on an adaptive strategy, which employs the aforementioned photovoltaic fault detection system based on an adaptive strategy. The method includes the following steps:

[0016] Data acquisition and preprocessing: The voltage, current, temperature and radiation intensity data of the photovoltaic module are collected in real time through a multi-sensor network, and dynamic noise reduction and standardization are performed using the Kalman filter algorithm;

[0017] Fault diagnosis and classification: Current and voltage curves, temperature changes and radiation intensity features are extracted from preprocessed data. Support vector machine and K-means clustering algorithm are used for fault mode recognition and classification. Fault development trend is predicted by regression algorithm.

[0018] Fault location impact analysis: Based on spatial distribution analysis and weighted shortest path algorithm, the fault source is located and the degree of impact of the fault on system performance is evaluated;

[0019] Adaptive maintenance decision-making: Based on the fault diagnosis and location results, dynamically adjust the operating parameters of equipment such as inverters, generate preventive maintenance plans, and automatically prioritize maintenance tasks according to the severity of the fault.

[0020] Furthermore, the data acquisition and preprocessing steps are implemented as follows:

[0021] Raw sensor data of voltage, current, temperature and radiation intensity of photovoltaic modules are collected in real time through a multi-sensor network.

[0022] The Kalman filter algorithm is used to remove noise and outliers from the original sensor data. The Kalman filter algorithm's data denoising formula is:

[0023] ;

[0024] in, The optimal estimate of the system state at time k is based on all observation data from time 1 to time k; Kalman gain; These are actual observed values; This is the observation matrix.

[0025] Standardization techniques are used to standardize and unify raw sensor data from different devices.

[0026] Anomaly detection algorithms are applied to monitor data in real time, and potential fault signs are identified by analyzing data trends to obtain preliminary abnormal data.

[0027] Furthermore, the fault diagnosis classification steps are implemented as follows:

[0028] Extract key features from preprocessed data;

[0029] The Support Vector Machine (SVM) machine learning algorithm is used to train the data and identify fault types. The function formula for fault classification using SVM is as follows:

[0030] ;

[0031] in, They are Lagrange multipliers; These are sample labels; These are training samples; It is the inner product operation, used to calculate the similarity of support vectors. Let be the multidimensional feature vector of the i-th sample, containing all the features of that sample. The sample data is from the training set; It is the bias term; N represents the total number of samples in the training set. During training, the support vector machine uses each sample to adjust the model's weights and biases in order to achieve optimal classification.

[0032] The K-means clustering algorithm is used to group different fault types according to their features;

[0033] The system uses regression algorithms to predict the possible future operating states of photovoltaic systems and monitor potential failure trends.

[0034] Furthermore, the implementation of the fault location impact analysis steps is as follows:

[0035] By analyzing the spatial distribution of sensor data and the feedback information from the inverter, the source of the fault can be accurately located.

[0036] After the source of the fault is identified, assess the status of each device and the overall impact of the fault on the photovoltaic system.

[0037] A thorough analysis of the photovoltaic panel array or individual modules is conducted to accurately identify and locate the location and cause of faults. A weighted shortest path algorithm is used for fault location; the path calculation formula for the weighted shortest path algorithm is as follows:

[0038] ;

[0039] in, It is the shortest path from node s to node . It is the set of all paths from s to t; It is the weight of each edge in the path;

[0040] By calculating and analyzing the redundancy design, the response of the redundancy mechanism can be evaluated, so as to develop a more effective recovery plan when some equipment fails.

[0041] Furthermore, the adaptive maintenance decision-making steps are implemented as follows:

[0042] Based on real-time data and fault prediction results, the operating parameters of the inverter and other key equipment are adjusted through adaptive control algorithms.

[0043] Based on the results of fault prediction and trend analysis, maintenance plans are developed in advance, and based on the equipment status and fault trends, decision-makers are helped to develop maintenance measures for impending faults.

[0044] For specified minor faults, the photovoltaic system can be automatically restored to normal operation through remote operation or adaptive strategies.

[0045] When multiple failures occur, maintenance tasks are prioritized by assessing the impact of the failures and the difficulty of recovery.

[0046] This invention employs the above technical solutions, utilizing multi-sensor data acquisition combined with Kalman filtering, adaptive anomaly detection, and data normalization processing to achieve high-precision, low-noise acquisition of key operational data such as voltage, current, temperature, and radiation, thereby significantly improving data reliability. Subsequently, through feature extraction, SVM fault identification, K-means classification, and trend prediction analysis, it achieves accurate identification and early warning of photovoltaic fault types, effectively reducing false alarm rates and enhancing fault prediction capabilities. The system combines spatial distribution analysis, weighted shortest path algorithm, module / array-level analysis, and redundancy assessment to achieve precise location of fault sources and quantitative assessment of their impact on overall system performance, improving fault location efficiency and maintenance targeting. Through adaptive control, preventative maintenance decision-making, automatic repair, and maintenance priority ranking functions, the system can dynamically adjust operating parameters based on real-time status and fault impact, formulate maintenance plans in advance, and prioritize the handling of critical faults, thereby reducing downtime, lowering operation and maintenance costs, and significantly improving the safety, stability, and long-term operating efficiency of the photovoltaic system. Attached Figure Description

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments;

[0048] Figure 1 This is a schematic diagram illustrating the principle structure of a photovoltaic fault detection system based on an adaptive strategy according to the present invention.

[0049] Figure 2 This is a flowchart illustrating a photovoltaic fault diagnosis method based on an adaptive strategy according to the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0051] like Figure 1 As shown, this invention discloses a specific embodiment of a photovoltaic fault detection system based on an adaptive strategy, including:

[0052] The data acquisition and preprocessing module is used to collect real-time data on voltage, current, temperature and radiation intensity of photovoltaic modules through a multi-sensor network, and to obtain preliminary abnormal data by dynamically denoising and standardizing the collected data through a Kalman filter algorithm.

[0053] The fault diagnosis and classification module is connected to the data acquisition and preprocessing module. It is used to extract features from the processed acquisition data and perform fault pattern recognition and classification on the extracted features based on support vector machine and K-means clustering algorithm. At the same time, it predicts fault trends and risks through regression algorithm.

[0054] The fault location and impact analysis module, connected to the fault diagnosis and classification module, is used to locate the fault location and its impact range through spatial distribution analysis and weighted shortest path algorithm; assess the impact of the fault on the overall performance of the photovoltaic system and provide repair strategies.

[0055] The adaptive maintenance decision module, connected to the fault location impact analysis module, is used to dynamically adjust equipment operating parameters based on fault diagnosis and location results to generate preventive maintenance strategies, and automatically prioritize maintenance based on fault severity for automated repair.

[0056] Furthermore, the photovoltaic data acquisition and preprocessing module includes a data acquisition sensor interface module, a data cleaning and denoising module, a data standardization and normalization module, and a real-time data monitoring and anomaly detection module. The data acquisition sensor interface module is responsible for acquiring real-time data from photovoltaic panels and inverter hardware to obtain raw sensor data. The data cleaning and denoising module removes noise and outliers from the sensor data using a Kalman filter adaptive filter to obtain denoised data. The data standardization and normalization module standardizes data from different acquisition devices to obtain data with a unified standard, ensuring that different data dimensions can be processed under the same standard. The real-time data monitoring and anomaly detection module uses anomaly detection algorithms to identify potential problems in the acquired data to obtain preliminary anomaly data.

[0057] Furthermore, the fault diagnosis and classification module includes a data feature extraction module, a fault mode recognition module, a classification and clustering module, and a fault prediction trend analysis module. The data feature extraction module extracts current and voltage curves, radiation changes, and temperature changes from the collected data. The fault mode recognition module uses SVM to train the collected data features and determine the fault type. The classification and clustering module classifies the fault types using the K-means algorithm. The fault prediction trend analysis module uses a regression algorithm to predict the future state of the photovoltaic system and monitor potential fault trends, further providing preventative maintenance support for the system.

[0058] Furthermore, the fault location impact analysis module includes a fault source location module, a system status assessment module, a module / array level analysis module, and a redundancy analysis module. The fault source location module locates the fault source by analyzing the spatial distribution of sensor data and inverter feedback information. The system status assessment module generates impact assessment results by monitoring the status of equipment to assess the impact of the fault. The module / array level analysis module analyzes the status of the photovoltaic array or individual modules to refine the fault location. The redundancy analysis module analyzes the system's redundancy design and calculates the redundancy mechanism response when a fault occurs, and evaluates the performance of the redundancy mechanism when a fault occurs to help formulate a repair strategy.

[0059] Furthermore, the adaptive control maintenance decision module includes an adaptive control algorithm module, a preventive maintenance decision support module, an automatic repair mechanism module, and a maintenance priority ranking module. The adaptive control algorithm module adjusts the operating parameters of the corresponding equipment (inverters and other equipment) in the photovoltaic system according to the real-time status of the photovoltaic system. The preventive maintenance decision support module formulates maintenance plans in advance based on fault prediction and trend analysis results. The automatic repair mechanism module automatically repairs minor faults of specified categories to restore system operation through remote operation or adaptive strategies. The maintenance priority ranking module prioritizes maintenance tasks by assessing the degree of fault impact and recovery difficulty.

[0060] In the above embodiments, the system utilizes multi-sensor data acquisition combined with Kalman filtering, adaptive anomaly detection, and data normalization to achieve high-precision, low-noise acquisition of key operational data such as voltage, current, temperature, and radiation, thereby significantly improving data reliability. Subsequently, through feature extraction, SVM fault identification, K-means classification, and trend prediction analysis, the system accurately identifies photovoltaic fault types and provides early warnings, effectively reducing false alarm rates and enhancing fault prediction capabilities. The system combines spatial distribution analysis, weighted shortest path algorithms, module / array-level analysis, and redundancy assessment to achieve precise location of fault sources and quantitative assessment of their impact on overall system performance, improving fault location efficiency and maintenance targeting. Through adaptive control, preventative maintenance decision-making, automatic repair, and maintenance priority ranking functions, the system can dynamically adjust operating parameters based on real-time status and fault impact, formulate maintenance plans in advance, and prioritize critical faults, thereby reducing downtime, lowering operation and maintenance costs, and significantly improving the safety, stability, and long-term operating efficiency of the photovoltaic system.

[0061] like Figure 2 As shown, this invention also discloses a photovoltaic fault diagnosis method based on an adaptive strategy, which employs the aforementioned photovoltaic fault detection system based on an adaptive strategy. The method includes the following steps:

[0062] Data acquisition and preprocessing: The voltage, current, temperature and radiation intensity data of the photovoltaic module are collected in real time through a multi-sensor network, and dynamic noise reduction and standardization are performed using the Kalman filter algorithm;

[0063] Fault diagnosis and classification: Current and voltage curves, temperature changes and radiation intensity features are extracted from preprocessed data. Support vector machine and K-means clustering algorithm are used for fault mode recognition and classification. Fault development trend is predicted by regression algorithm.

[0064] Fault location impact analysis: Based on spatial distribution analysis and weighted shortest path algorithm, the fault source is located and the degree of impact of the fault on system performance is evaluated;

[0065] Adaptive maintenance decision-making: Based on the fault diagnosis and location results, dynamically adjust the operating parameters of equipment such as inverters, generate preventive maintenance plans, and automatically prioritize maintenance tasks according to the severity of the fault.

[0066] Furthermore, the data acquisition and preprocessing steps are implemented as follows:

[0067] (From hardware devices such as photovoltaic panels and inverters) Raw sensor data on voltage, current, temperature, and radiation intensity of photovoltaic modules are collected in real time through a multi-sensor network, serving as the basis for fault diagnosis and subsequent analysis;

[0068] The Kalman filter algorithm is used to remove noise and outliers from the original sensor data. The Kalman filter algorithm's data denoising formula is:

[0069] ;

[0070] in, This represents the best estimate of the system state at time k; Kalman gain; These are actual observed values; This is the observation matrix.

[0071] Standardization techniques are used to standardize and unify raw sensor data from different devices.

[0072] Anomaly detection algorithms are applied to monitor data in real time, and potential fault signs are identified by analyzing data trends to obtain preliminary abnormal data.

[0073] Furthermore, the fault diagnosis classification steps are implemented as follows:

[0074] Extract key features from preprocessed data;

[0075] The Support Vector Machine (SVM) machine learning algorithm is used to train the data and identify fault types. The function formula for fault classification using SVM is as follows:

[0076] ;

[0077] in, They are Lagrange multipliers; These are sample labels; These are training samples; It is the inner product operation, used to calculate the similarity of support vectors. Let be the multidimensional feature vector of the i-th sample, containing all the features of that sample. The sample data is from the training set; is the bias term; N represents the total number of samples in the training set. During training, the support vector machine uses each sample to adjust the model's weights and biases in order to achieve optimal classification; the K-means clustering algorithm is used to group different fault types according to features.

[0078] The system uses regression algorithms to predict the possible future operating states of photovoltaic systems and monitor potential failure trends.

[0079] Furthermore, the implementation of the fault location impact analysis steps is as follows:

[0080] By analyzing the spatial distribution of sensor data and the feedback information from the inverter, the source of the fault can be accurately located.

[0081] After the source of the fault is identified, assess the status of each device and the overall impact of the fault on the photovoltaic system.

[0082] A thorough analysis of the photovoltaic panel array or individual modules is conducted to accurately identify and locate the location and cause of faults. A weighted shortest path algorithm is used for fault location; the path calculation formula for the weighted shortest path algorithm is as follows:

[0083] ;

[0084] in, It is the shortest path from node s to node . It is the set of all paths from s to t; It is the weight of each edge in the path;

[0085] By calculating and analyzing the redundancy design, the response of the redundancy mechanism can be evaluated, so as to develop a more effective recovery plan when some equipment fails.

[0086] Furthermore, the adaptive maintenance decision-making steps are implemented as follows:

[0087] Based on real-time data and fault prediction results, the operating parameters of the inverter and other key equipment are adjusted through adaptive control algorithms.

[0088] Based on the results of fault prediction and trend analysis, maintenance plans are developed in advance, and based on the equipment status and fault trends, decision-makers are helped to develop maintenance measures for impending faults.

[0089] For specified minor faults, the photovoltaic system can be automatically restored to normal operation through remote operation or adaptive strategies.

[0090] When multiple failures occur, maintenance tasks are prioritized by assessing the impact of the failures and the difficulty of recovery, so that maintenance can be performed in sequence according to priority.

[0091] In the above embodiments, a high-efficiency data acquisition module acquires real-time sensor data such as current, voltage, and temperature from the photovoltaic panel and inverter. A Kalman filter algorithm is used to remove noise and outliers, ensuring data accuracy. After data preprocessing, an anomaly detection algorithm is applied to monitor the data in real time, promptly identifying potential fault signs. Through machine learning algorithms, the system can accurately identify and classify fault modes, predict fault trends and risks, and precisely locate fault sources and their impact range by analyzing the spatial distribution of fault sources and equipment status. The response of redundant designs is evaluated, and effective repair and recovery plans are formulated. These measures collectively ensure the stability and efficient operation of the photovoltaic system and effectively reduce the impact of faults on the system.

[0092] Working Principle: When in use, this invention relies on a highly efficient photovoltaic data acquisition and preprocessing module. It acquires key data such as current, voltage, temperature, and radiation intensity from photovoltaic panels, inverters, and other devices in real time. These sensor data are processed using a Kalman filter algorithm to remove noise and outliers, ensuring the accuracy and reliability of the data. Through data standardization processing, data collected from different devices are processed uniformly to facilitate subsequent anomaly detection and fault analysis. The system also monitors real-time data using anomaly detection algorithms to promptly identify potential fault signs, providing a foundation for subsequent diagnosis and location.

[0093] After data preprocessing, the system enters the fault diagnosis stage. The data feature extraction module extracts key fault features, such as current and voltage curves, temperature and radiation changes. The fault mode recognition module uses support vector machine (SVM) to train these feature data to accurately determine whether there is a fault in the photovoltaic system. The K-means clustering algorithm is used to classify different fault types. This process enables the system to accurately identify different types of faults and provide basic data for subsequent fault prediction. The regression algorithm is also used for fault trend prediction, providing early warning of potential faults and helping the system to perform preventive maintenance.

[0094] After fault diagnosis, the fault location impact analysis module further analyzes the specific location of the fault source and its impact on the photovoltaic system. By analyzing the spatial distribution of sensor data and combining it with inverter feedback information, the fault source is accurately located. By assessing the status of each device in the system and the impact after the fault occurs, the system can quantify the degree of impact of the fault on the overall photovoltaic system and provide precise repair strategies. The system also performs more detailed analysis on photovoltaic panel arrays or individual modules to ensure the accurate location of the fault source. The redundancy analysis module analyzes the response mechanism of the redundancy design and evaluates the emergency response capability of the system when some devices fail, thereby helping to formulate an effective recovery plan.

[0095] After the source and scope of system failure are determined, the system will adjust the operating parameters of the equipment in the photovoltaic system in real time based on the adaptive control maintenance decision module to ensure the stable operation of the photovoltaic system after the failure occurs. By optimizing the operating parameters of inverters and other equipment, maintenance plans are formulated in advance, and automatic repair mechanisms are used to handle failures. For small-scale failures, the system can automatically restore system operation through remote operation or adaptive strategies, which greatly reduces downtime and maintenance costs.

[0096] This invention enables efficient data acquisition, accurate fault diagnosis, precise fault location, and flexible adaptive maintenance. Through comprehensive fault detection and prediction, the system can not only promptly detect and diagnose faults, but also dynamically adjust according to the real-time status of the system, improving the stability, efficiency, and long-term operational capability of the photovoltaic system, and significantly reducing the impact of faults on the system and operation and maintenance costs.

[0097] This invention employs the above technical solutions, utilizing multi-sensor data acquisition combined with Kalman filtering, adaptive anomaly detection, and data normalization processing to achieve high-precision, low-noise acquisition of key operational data such as voltage, current, temperature, and radiation, thereby significantly improving data reliability. Subsequently, through feature extraction, SVM fault identification, K-means classification, and trend prediction analysis, it achieves accurate identification and early warning of photovoltaic fault types, effectively reducing false alarm rates and enhancing fault prediction capabilities. The system combines spatial distribution analysis, weighted shortest path algorithm, module / array-level analysis, and redundancy assessment to achieve precise location of fault sources and quantitative assessment of their impact on overall system performance, improving fault location efficiency and maintenance targeting. Through adaptive control, preventative maintenance decision-making, automatic repair, and maintenance priority ranking functions, the system can dynamically adjust operating parameters based on real-time status and fault impact, formulate maintenance plans in advance, and prioritize the handling of critical faults, thereby reducing downtime, lowering operation and maintenance costs, and significantly improving the safety, stability, and long-term operating efficiency of the photovoltaic system.

[0098] Obviously, the described embodiments are only a portion, not all, of the embodiments of this application. Without conflict, the embodiments and features described and illustrated herein can be combined with each other. The components of the embodiments of this application generally described and illustrated in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

Claims

1. A photovoltaic fault detection system based on an adaptive strategy, characterized in that: include: The data acquisition and preprocessing module is used to collect real-time data on voltage, current, temperature and radiation intensity of photovoltaic modules through a multi-sensor network, and to obtain preliminary abnormal data by dynamically denoising and standardizing the collected data through a Kalman filter algorithm. The fault diagnosis and classification module is connected to the data acquisition and preprocessing module. It is used to extract features from the processed acquisition data and perform fault pattern recognition and classification on the extracted features based on support vector machine and K-means clustering algorithm. At the same time, it predicts fault trends and risks through regression algorithm. The fault location and impact analysis module, connected to the fault diagnosis and classification module, is used to locate the fault location and its impact range through spatial distribution analysis and weighted shortest path algorithm; assess the impact of the fault on the overall performance of the photovoltaic system and provide repair strategies. The adaptive maintenance decision module, connected to the fault location impact analysis module, is used to dynamically adjust equipment operating parameters based on fault diagnosis and location results to generate preventive maintenance strategies, and automatically prioritize maintenance based on fault severity for automated repair.

2. The photovoltaic fault detection system based on an adaptive strategy according to claim 1, characterized in that: The photovoltaic data acquisition and preprocessing module includes a data acquisition sensor interface module, a data cleaning and denoising module, a data standardization module, and a real-time data monitoring and anomaly detection module. The data acquisition sensor interface module is responsible for acquiring real-time data from photovoltaic panels and inverter hardware devices to obtain raw sensor data. The data cleaning and denoising module is used to remove noise and outliers from the sensor data using a Kalman filter adaptive filter to obtain denoised data. The data standardization module is used to standardize data from different acquisition devices to obtain data with a unified standard, ensuring that different data dimensions can be processed under the same standard; the real-time data monitoring anomaly detection module is used to identify potential problems in the acquired data using anomaly detection algorithms to obtain preliminary abnormal data.

3. The photovoltaic fault detection system based on an adaptive strategy according to claim 1, characterized in that: The fault diagnosis and classification module includes a data feature extraction module, a fault mode recognition module, a classification and clustering module, and a fault prediction trend analysis module; the data feature extraction module extracts the features of the collected data, such as current and voltage curves, radiation changes, and temperature changes, from the collected data. The fault mode recognition module uses SVM to train the features of the collected data and determine the fault type; The classification and clustering module classifies fault types using the K-means algorithm; the fault prediction and trend analysis module predicts the future state of the photovoltaic system and monitors potential fault trends using a regression algorithm.

4. The photovoltaic fault detection system based on an adaptive strategy according to claim 1, characterized in that: The fault location impact analysis module includes a fault source location module, a system status assessment module, a module / array-level analysis module, and a redundancy analysis module. The fault source location module locates the fault source by analyzing the spatial distribution of sensor data and inverter feedback information. The system status assessment module generates impact assessment results by monitoring the status of equipment to assess the impact of the fault. The module / array-level analysis module analyzes the status of the photovoltaic array or individual modules to refine the fault location. The redundancy analysis module analyzes the system's redundancy design and calculates the redundancy mechanism response when a fault occurs, and evaluates the performance of the redundancy mechanism when a fault occurs to help formulate a repair strategy.

5. The photovoltaic fault detection system based on an adaptive strategy according to claim 1, characterized in that: The adaptive control maintenance decision module includes an adaptive control algorithm module, a preventive maintenance decision support module, an automatic repair mechanism module, and a maintenance priority ranking module. The adaptive control algorithm module adjusts the operating parameters of the corresponding equipment in the photovoltaic system according to the real-time status of the photovoltaic system. The preventive maintenance decision support module formulates maintenance plans in advance based on fault prediction and trend analysis results. The automatic repair mechanism module automatically repairs minor faults of specified categories to restore system operation through remote operation or adaptive strategies. The maintenance priority ranking module prioritizes maintenance tasks by assessing the impact of faults and the difficulty of recovery.

6. A photovoltaic fault diagnosis method based on an adaptive strategy, employing a photovoltaic fault detection system based on an adaptive strategy as described in any one of claims 1 to 5, characterized in that: The method includes the following steps: Data acquisition and preprocessing: The voltage, current, temperature and radiation intensity data of the photovoltaic module are collected in real time through a multi-sensor network, and dynamic noise reduction and standardization are performed using the Kalman filter algorithm; Fault diagnosis and classification: Current and voltage curves, temperature changes and radiation intensity features are extracted from preprocessed data. Support vector machine and K-means clustering algorithm are used for fault mode recognition and classification. Fault development trend is predicted by regression algorithm. Fault location impact analysis: Based on spatial distribution analysis and weighted shortest path algorithm, the fault source is located and the degree of impact of the fault on system performance is evaluated; Adaptive maintenance decision-making: Based on the fault diagnosis and location results, dynamically adjust the operating parameters of equipment such as inverters, generate preventive maintenance plans, and automatically prioritize maintenance tasks according to the severity of the fault.

7. The photovoltaic fault diagnosis method based on an adaptive strategy according to claim 6, characterized in that: The data acquisition and preprocessing steps are implemented as follows: Raw sensor data of voltage, current, temperature and radiation intensity of photovoltaic modules are collected in real time through a multi-sensor network. The Kalman filter algorithm is used to remove noise and outliers from the original sensor data. The Kalman filter algorithm's data denoising formula is: ； in, This represents the best estimate of the system state at time k; Kalman gain; These are actual observed values; The observation matrix; Standardization techniques are used to standardize and unify raw sensor data from different devices. Anomaly detection algorithms are applied to monitor data in real time, and potential fault signs are identified by analyzing data trends to obtain preliminary abnormal data.

8. The photovoltaic fault diagnosis method based on an adaptive strategy according to claim 6, characterized in that: The fault diagnosis and classification steps are implemented as follows: Extract key features from preprocessed data; The Support Vector Machine (SVM) machine learning algorithm is used to train the data and identify fault types. The function formula for fault classification using SVM is: ； in, They are Lagrange multipliers; These are sample labels; These are training samples; It is the inner product operation, used to calculate the similarity of support vectors. Let be the multidimensional feature vector of the i-th sample. The sample data is from the training set; It is the bias term; N represents the total number of samples in the training set; The K-means clustering algorithm is used to group different fault types according to their features; The system uses regression algorithms to predict the possible future operating states of photovoltaic systems and monitor potential failure trends.

9. The photovoltaic fault detection system and diagnosis method based on an adaptive strategy according to claim 6, characterized in that: The steps for fault location impact analysis are as follows: By analyzing the spatial distribution of sensor data and the feedback information from the inverter, the source of the fault can be accurately located. After the source of the fault is identified, assess the status of each device and the overall impact of the fault on the photovoltaic system. A thorough analysis of the photovoltaic panel array or individual modules is conducted to accurately identify and locate the location and cause of faults. A weighted shortest path algorithm is used for fault location; the path calculation formula for the weighted shortest path algorithm is as follows: ； in, It is the shortest path from node s to node . It is the set of all paths from s to t; It is the weight of each edge in the path; By calculating and analyzing the redundancy design, the response of the redundancy mechanism can be evaluated, so as to develop a more effective recovery plan when some equipment fails.

10. The photovoltaic fault detection system and diagnosis method based on an adaptive strategy according to claim 6, characterized in that: The adaptive maintenance decision-making steps are implemented as follows: Based on real-time data and fault prediction results, the operating parameters of the inverter and other key equipment are adjusted through adaptive control algorithms. Based on the results of fault prediction and trend analysis, maintenance plans are developed in advance, and based on the equipment status and fault trends, decision-makers are helped to develop maintenance measures for impending faults. For specified minor faults, the photovoltaic system can be automatically restored to normal operation through remote operation or adaptive strategies. When multiple failures occur, maintenance tasks are prioritized by assessing the impact of the failures and the difficulty of recovery.

Images