Photovoltaic panel fault diagnosis method, device, equipment and storage medium

Abstract

The application discloses a photovoltaic panel fault diagnosis method, device and equipment and a storage medium, and comprises the following steps: collecting multi-source operation parameters of a photovoltaic panel, pre-processing the multi-source operation parameters through an edge computing gateway to obtain real-time operation parameters after preprocessing; fusing the real-time operation parameters after preprocessing and historical fault data, inputting the fusion result into a preset photovoltaic panel health assessment model to generate a health state assessment result; based on the health state assessment result, determining fault type information and fault positioning information of the photovoltaic panel through a feature analysis algorithm; and performing space-time alignment on the fault type information and the fault positioning information to generate a fault distribution diagram of the photovoltaic panel. The application pre-processes multi-source operation parameters through an edge computing gateway, fuses real-time operation parameters and historical fault data, and determines fault type information and fault positioning information through a feature analysis algorithm, thereby improving the accuracy and reliability of photovoltaic panel fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, equipment and storage medium for diagnosing photovoltaic panel faults. Background Technology

[0002] With the rapid growth of global photovoltaic power plant installed capacity, the long-term reliable operation of photovoltaic modules has become a core element in ensuring the profitability of power plants. In actual operation, photovoltaic panels can experience a decline in power generation efficiency and even lead to safety accidents such as fires due to various failure modes, including hot spots, microcracks, PID (potential-induced degradation), and dust accumulation. Therefore, timely and accurate identification of failure types and location is of great significance for reducing operation and maintenance costs and increasing power generation.

[0003] Traditional fault diagnosis typically relies on threshold judgment methods based on string-level or inverter-level electrical parameters (voltage, current, power). This method depends on a single data source, is easily affected by irradiance, temperature variations, and grid fluctuations, has limited fault location accuracy, and cannot distinguish specific fault types. Consequently, the accuracy and reliability of photovoltaic panel fault diagnosis are low, failing to provide a reliable basis for subsequent precise operation and maintenance decisions.

[0004] Therefore, there is an urgent need for a photovoltaic panel fault diagnosis method that can improve the accuracy and reliability of photovoltaic panel fault diagnosis. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for diagnosing photovoltaic panel faults, aiming to solve the technical problem of low accuracy and reliability in existing photovoltaic panel fault diagnosis technologies.

[0006] To achieve the above objectives, the present invention provides a photovoltaic panel fault diagnosis method, the method comprising the following steps: Multi-source operating parameters of photovoltaic panels are collected, and the multi-source operating parameters are preprocessed through an edge computing gateway to obtain preprocessed real-time operating parameters; The preprocessed real-time operating parameters and historical fault data are fused together, and the fusion result is input into a preset photovoltaic panel health assessment model to generate a health status assessment result. Based on the health status assessment results, the fault type information and fault location information of the photovoltaic panel are determined by feature parsing algorithm; The fault type information and the fault location information are spatiotemporally aligned to generate a fault distribution map of the photovoltaic panel.

[0007] Optionally, the step of determining the fault type information and fault location information of the photovoltaic panel based on the health status assessment results using a feature parsing algorithm includes: The key features in the health status assessment results are analyzed using a feature parsing algorithm to obtain the analysis results; The analysis results are matched with a preset fault knowledge base, and the fault type information of the photovoltaic panel is determined based on the matching results. A distributed inspection task is triggered based on the fault type information, and the fault location information returned by the distributed inspection task is received.

[0008] Optionally, the step of triggering a distributed inspection task based on the fault type information and receiving fault location information returned by the distributed inspection task includes: Based on the fault type information, a distributed inspection task is created in the task scheduling queue, and the distributed inspection task is sent to the corresponding UAV. The UAV acquires infrared-visible dual-mode images, and the dual-mode images are input into a lightweight generative network to obtain defect mask images. The defect mask image is bound to the BeiDou positioning data transmitted back by the UAV in real time to generate defect pixel blocks with geographic coordinates. The defective pixel blocks are clustered and merged to obtain fault location information.

[0009] Optionally, after the step of spatiotemporally aligning the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel, the method further includes: Based on the fault type information corresponding to the fault category label, query the operation and maintenance strategy library for the handling strategy that matches the fault category label; Based on the handling strategy and the fault location information, an operation and maintenance instruction is generated and sent to the corresponding operation and maintenance terminal. After receiving the maintenance instruction, the maintenance terminal performs corresponding maintenance operations on the photovoltaic panel according to the maintenance instruction.

[0010] Optionally, the method further includes: When the fault type information corresponds to the fault category label of the photovoltaic panel having a dust accumulation fault or a shading fault, a cleaning instruction is generated based on the coordinates of the shading area in the fault distribution map; The cleaning instructions control an adaptive cleaning robot cluster to plan the cleaning path of the photovoltaic panel using a reinforcement learning algorithm, and collect humidity sensor signals in real time during the cleaning process. The cleaning path is adjusted based on the humidity sensor signal to obtain the adjusted cleaning path.

[0011] Optionally, after the step of spatiotemporally aligning the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel, the method further includes: Historical fault records and the health status assessment results are input into the digital twin platform, and the degradation process of each component in the photovoltaic panel is simulated through the digital twin platform to obtain simulation results; The simulation results are compared with a preset degradation threshold to obtain a comparison result; When the comparison result meets the warning triggering conditions, a warning message is generated and pushed to the operation and maintenance management platform.

[0012] Optionally, the step of fusing the preprocessed real-time operating parameters and historical fault data, and inputting the fusion result into a preset photovoltaic panel health assessment model to generate a health status assessment result includes: The preprocessed real-time operating parameters are converted into time-series feature vectors, and the historical fault data is mapped into fault prior weights. The prior fault weights and the temporal feature vectors are weighted and fused using a transfer learning mechanism to obtain a fusion result. The fusion results are input into a preset photovoltaic panel health assessment model to generate a health status assessment result.

[0013] Furthermore, to achieve the above objectives, the present invention also proposes a photovoltaic panel fault diagnosis device, the device comprising: The parameter acquisition module is used to acquire multi-source operating parameters of the photovoltaic panel and preprocess the multi-source operating parameters through the edge computing gateway to obtain the preprocessed real-time operating parameters. The status assessment module is used to fuse the preprocessed real-time operating parameters and historical fault data, and input the fusion result into the preset photovoltaic panel health assessment model to generate a health status assessment result. The fault determination module is used to determine the fault type information and fault location information of the photovoltaic panel based on the health status assessment results and through a feature parsing algorithm. The information processing module is used to perform spatiotemporal alignment of the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel.

[0014] Furthermore, to achieve the above objectives, the present invention also proposes a photovoltaic panel fault diagnosis device, the device comprising: a memory, a processor, and a photovoltaic panel fault diagnosis program stored in the memory and executable on the processor, the photovoltaic panel fault diagnosis program being configured to implement the steps of the photovoltaic panel fault diagnosis method described above.

[0015] In addition, to achieve the above objectives, the present invention also proposes a storage medium storing a photovoltaic panel fault diagnosis program, wherein when the photovoltaic panel fault diagnosis program is executed by a processor, it implements the steps of the photovoltaic panel fault diagnosis method described above.

[0016] This invention discloses a method for collecting multi-source operating parameters of photovoltaic (PV) panels and preprocessing these parameters using an edge computing gateway to obtain preprocessed real-time operating parameters. The preprocessed real-time operating parameters are then fused with historical fault data, and the fusion result is input into a pre-defined PV panel health assessment model to generate a health status assessment result. Based on the health status assessment result, a feature parsing algorithm is used to determine the fault type and fault location information of the PV panel. Finally, the fault type and fault location information are spatiotemporally aligned to generate a fault distribution map of the PV panel. Because this invention preprocesses the multi-source operating parameters using an edge computing gateway to obtain preprocessed real-time operating parameters, fuses these parameters with historical fault data, inputs the fused data into a pre-defined PV panel health assessment model, and then uses a feature parsing algorithm to determine the fault type and fault location information and generate a fault distribution map, this invention improves the accuracy and reliability of PV panel fault diagnosis compared to existing technologies. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the photovoltaic panel fault diagnosis method of the present invention; Figure 2 This is a flowchart illustrating the second embodiment of the photovoltaic panel fault diagnosis method of the present invention; Figure 3 This is a flowchart illustrating the third embodiment of the photovoltaic panel fault diagnosis method of the present invention; Figure 4 This is a structural block diagram of the first embodiment of the photovoltaic panel fault diagnosis device of the present invention; Figure 5 This is a schematic diagram of the structure of a photovoltaic panel fault diagnosis device in the hardware operating environment involved in the embodiments of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0020] This invention provides a method for diagnosing photovoltaic panel faults, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the photovoltaic panel fault diagnosis method of the present invention.

[0021] In this embodiment, the photovoltaic panel fault diagnosis method includes steps S10 to S40: Step S10: Collect multi-source operating parameters of the photovoltaic panel, and preprocess the multi-source operating parameters through the edge computing gateway to obtain the preprocessed real-time operating parameters.

[0022] It should be noted that the executing entity in this embodiment can be a computer server device with data processing, network communication, and program execution functions applied in photovoltaic panel fault diagnosis scenarios, such as a server, tablet computer, or personal computer, or an electronic device capable of performing the above functions (such as photovoltaic panel fault diagnosis equipment). The following uses an operation and maintenance system including photovoltaic panel fault diagnosis equipment (hereinafter referred to as the system) as an example to illustrate this embodiment and the following embodiments.

[0023] It should be understood that a photovoltaic panel can refer to a plate-shaped power generation unit that uses the semiconductor photovoltaic effect to directly convert solar radiation energy into electrical energy. It is the smallest independent power generation unit of a photovoltaic power station and is usually composed of several solar cells, encapsulated glass, backsheet and junction box.

[0024] Understandably, multi-source operating parameters can refer to operating condition quantities derived from different types of sensors, including at least the following parameters: electrical quantities (current, voltage, and power of the photovoltaic panel, etc.), physical quantities (surface temperature and backsheet temperature of the photovoltaic panel, etc.), meteorological quantities (irradiance, ambient temperature, humidity, and wind speed, etc.), and image quantities (grayscale values ​​of infrared thermal imaging and visible light imaging). These parameters, after being timestamped, collectively constitute the original dataset used for fault diagnosis, i.e., multi-source operating parameters.

[0025] In practical implementation, temperature distribution data and abnormal current and voltage data of photovoltaic panels can be monitored through distributed fiber optic sensor arrays; irradiance change data and ambient humidity data can be collected through micro weather stations; and temperature distribution data, abnormal current and voltage data, irradiance change data, and ambient humidity data can be used as multi-source operating parameters of photovoltaic panels.

[0026] It should be noted that the edge computing gateway can be an industrial-grade computing and communication node deployed near the photovoltaic array. It has the capabilities of local data acquisition, caching, protocol conversion, filtering and noise reduction, feature extraction and model inference. It can complete data preprocessing in milliseconds and upload the compressed high-value information to the cloud or backend system, which can effectively reduce transmission bandwidth and cloud computing load.

[0027] In practical implementation, noise filtering and data standardization of multi-source operating parameters can be performed through edge computing gateways. Then, feature parameters related to the health status of photovoltaic panels can be extracted from the processed data to obtain preprocessed real-time operating parameters.

[0028] Step S20: The preprocessed real-time operating parameters and historical fault data are fused together, and the fusion result is input into the preset photovoltaic panel health assessment model to generate a health status assessment result.

[0029] It should be noted that the preset photovoltaic panel health assessment model can be a model built based on a Transformer network. The health status assessment result can be a probability vector output by the model after one forward inference. Each dimension of the probability vector corresponds to a fault category, and the dimension value is an estimate of the probability of occurrence of that type of fault. This probability vector serves as the input to the subsequent feature parsing algorithm to derive fault type information and fault location information.

[0030] In the specific implementation, multi-source historical operating parameters of photovoltaic panels and historical fault data can be collected. Then, the multi-source historical operating parameters and historical fault data are preprocessed (e.g., data cleaning and data normalization). An initial photovoltaic panel health assessment model is constructed based on a Transformer-based deep learning model. The initial photovoltaic panel health assessment model is trained based on the preprocessed historical fault data and preprocessed multi-source historical operating parameters to obtain a preset photovoltaic panel health assessment model.

[0031] It should be understood that multi-source historical operating parameters include electrical parameters (such as current and voltage), thermal parameters (such as temperature), and environmental parameters (such as irradiance and humidity). Historical fault data includes information such as fault type, fault occurrence time, and fault location. This preset photovoltaic panel health assessment model can output a probability vector representing the probability of the photovoltaic panel being in different health states. By analyzing the probability vector, the current health state of the photovoltaic panel can be determined, and further fault type information and fault location information can be generated.

[0032] Understandably, fusing preprocessed real-time operating parameters and historical fault data, and then inputting the fusion result into a preset photovoltaic panel health assessment model, improves the reliability of the health status assessment results.

[0033] In a specific implementation, the preprocessed real-time operating parameters can be converted into time-series feature vectors, and the historical fault data can be mapped into fault prior weights. The fault prior weights and the time-series feature vectors can be weighted and fused through a transfer learning mechanism to obtain a fusion result. The fusion result is then input into a preset photovoltaic panel health assessment model to generate a health status assessment result.

[0034] It should be explained that the time series feature vector can refer to a one-dimensional or multi-dimensional numerical sequence extracted from the preprocessed real-time operating parameters according to a preset time window (such as a 5-minute sliding window), and formed into a fixed-length vector after normalization, detrending, Fourier or wavelet transform, and statistical calculation. It is used to characterize the dynamic operating behavior of the photovoltaic panel during that time period.

[0035] It should be noted that the prior weights of the faults can be a vector of feature dimension weight coefficients that are strongly correlated with various faults and obtained from historical fault data statistics. Each component corresponds to the probability of occurrence or contribution of a fault mode on the corresponding feature.

[0036] Understandably, by using prior fault weights to weight the time-series feature vectors, the subsequent preset photovoltaic panel health assessment model prioritizes historical high-probability fault dimensions, significantly reducing the fault miss rate.

[0037] It should be understood that by using the transfer learning mechanism to weight and fuse the prior fault weights with the temporal feature vectors, a fusion result is obtained, which simultaneously includes real-time dynamic features and historical experience information, making the health status assessment results more accurate.

[0038] Furthermore, through the transfer learning mechanism, historical fault knowledge can be transferred to the current scenario. It can quickly adapt to different power plants and different component models without the need to collect a large number of new fault samples, thereby improving the generalization ability and diagnostic accuracy of the preset photovoltaic panel health assessment model.

[0039] Step S30: Based on the health status assessment results, determine the fault type information and fault location information of the photovoltaic panel through a feature parsing algorithm.

[0040] It is understood that the above feature parsing algorithm can be the SHAP (SHapley Additive exPlanations) algorithm or other feature parsing algorithms, and this embodiment does not limit it.

[0041] In a specific implementation, a feature parsing algorithm can be used to parse the key features in the health status assessment results to obtain the parsing results; the parsing results are matched with a preset fault knowledge base, and the fault type information of the photovoltaic panel is determined based on the matching results; a distributed inspection task is triggered based on the fault type information, and the fault location information returned by the distributed inspection task is received.

[0042] It should be understood that key features can be a subset of features whose contribution exceeds a set threshold after calculating the SHAP value of the health status assessment results. These features are strongly correlated with the physical mechanism of a certain type of fault (such as local temperature rise, current drop, and increased series resistance) and are regarded as the decisive basis for determining the fault type.

[0043] The pre-built fault knowledge base can be a data table or graph built and stored offline. Each record can contain a mapping relationship of "key feature combination - fault type - confidence level" (e.g., abnormal increase in local temperature, abnormal change in current and voltage - hot spot effect - 0.8). It can be used in the online stage to match the key feature subset in the analysis results with the records in the database, and determine the fault type information of the photovoltaic panel based on the matching results.

[0044] It should be noted that the key feature combination refers to a set of features that can uniquely or highly correlatedly identify a certain type of fault; the fault type refers to the specific fault situation that may occur in the photovoltaic panel (e.g., hot spot effect, dust accumulation, microcracks, PID effect, and shading); the confidence level refers to the degree of confidence of the model in judging a certain fault type, which is usually a value between 0 and 1. For example, a hot spot effect of -0.8 means that there is an 80% confidence level that the current fault is a hot spot effect.

[0045] It should be explained that distributed inspection tasks can be executable drone operation instruction packages that are automatically generated based on fault type information. These tasks not only include visual inspection of the photovoltaic panel surface, but can also be combined with multi-source operating parameters for comprehensive analysis, thereby achieving more accurate fault diagnosis and location.

[0046] It should be noted that the step of triggering a distributed inspection task based on the fault type information and receiving the fault location information returned by the distributed inspection task includes: creating a distributed inspection task in the task scheduling queue according to the fault type information and sending the distributed inspection task to the corresponding UAV; acquiring infrared-visible dual-mode images through the UAV and inputting the dual-mode images into a lightweight generative network to obtain a defect mask image; binding the defect mask image with the BeiDou positioning data transmitted back by the UAV in real time to generate defect pixel blocks with geographic coordinates; and clustering and merging the defect pixel blocks to obtain fault location information.

[0047] It should be explained that the task scheduling queue can be a priority-sorting cache maintained in the system, used to temporarily store distributed inspection tasks to be executed. The task scheduling queue is automatically sorted according to the urgency of the fault, the drone's battery level, and airspace control constraints. According to the order of this task scheduling queue, the system assigns distributed inspection tasks to appropriate drones for execution in sequence, realizing the orderly and efficient allocation and load balancing of distributed inspection tasks.

[0048] Understandably, an infrared-visible dual-mode image can refer to two different modalities of image data obtained by simultaneously equipping an infrared thermal imaging camera and a visible light camera on a drone to capture images of the same target area, and then fusing these two images (i.e., aligning the two images at the pixel level) to form a dual-mode image.

[0049] It should be understood that visible light images can reflect the appearance, texture, color, and other characteristics of a target. Infrared thermal imaging images can reflect the temperature distribution characteristics of a target. When a photovoltaic panel malfunctions (such as hot spots or microcracks), the temperature will be abnormal, which will be clearly shown in the infrared image. Therefore, fusing or jointly analyzing visible light images and infrared thermal imaging images can simultaneously obtain information about the appearance and thermal characteristics of the fault, further improving the accuracy of fault diagnosis.

[0050] It should be explained that lightweight generative networks can be generative models that have undergone channel pruning, knowledge distillation, or depthwise separable convolution compression (such as lightweight GAN networks). They can be deployed on UAV onboard chips or edge gateways and can complete inference in the 100ms range. They can take a dual-mode image as input and output a defect mask image, ensuring real-time fault diagnosis.

[0051] It should be understood that the defect mask image can be a single-channel binary or probabilistic map output by a lightweight generative network, where the defect pixel value is 1 (or 255) and the background pixel value is 0. It is used to accurately identify the spatial shape and size of the fault area in the infrared-visible dual-mode image for subsequent geomapping.

[0052] Furthermore, a defect pixel block with geographic coordinates can refer to the mapping of the coordinates of each connected pixel in the defect mask image to the Earth coordinate system using collinear equations or SLAM algorithms, based on the BeiDou positioning data (latitude, longitude, elevation, heading angle) transmitted back by the UAV in real time and the camera's intrinsic and extrinsic parameters model. This generates a geographic block with a one-to-one correspondence between "defect pixel - latitude and longitude", thereby achieving precise fault location.

[0053] In practical implementation, defective pixel blocks that are less than a set threshold (e.g., 1 meter) in the geographic coordinate system and have the same fault type can be clustered and merged using a clustering algorithm (e.g., DBSCAN algorithm or K-means++ algorithm) to obtain fault location information. This eliminates duplicate counting caused by image segmentation fragments and improves the accuracy of fault location.

[0054] Step S40: Spatiotemporally align the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel.

[0055] It should be noted that spatiotemporal alignment refers to synchronizing and matching fault type information and fault location information in both time and space dimensions, that is, achieving temporal and spatial alignment to ensure the consistency of this information in time and the accuracy in space. Specifically, the purpose of spatiotemporal alignment is to integrate fault type information (such as hot spot effect, PID effect, etc.) with fault location information (such as the specific location coordinates of the fault) to generate a comprehensive fault distribution map.

[0056] Furthermore, time alignment ensures the consistency of fault type information and fault location information in time. The timestamps of fault type information and fault location information can be aligned. If the timestamps are inconsistent, they can be adjusted by interpolation or time window matching, thereby achieving time alignment of fault type information and fault location information.

[0057] Furthermore, spatial alignment ensures the spatial accuracy of fault type and fault location information. Spatial coordinates can be extracted from the fault type information, which may be the fault location obtained through distributed fiber optic sensor arrays or infrared thermal imaging images; spatial coordinates can be extracted from the fault location information, which are typically generated by UAV inspection missions and obtained through the BeiDou positioning system as precise geographic coordinates. By transforming these two different sources of spatial coordinates into the same coordinate system, spatial alignment of fault type and fault location information is achieved.

[0058] In the specific implementation, fault type information within the same time interval and spatial range can be associated and matched based on the timestamp and geographical coordinates contained in the fault location information; a unique identifier and corresponding visualization style are assigned to each fault type according to the fault type information; the matched fault information is overlaid and fused with the three-dimensional model of the power station, and rendered and marked at the corresponding geographical coordinates according to the visualization style; based on the rendering and marking results, a visualization map is generated as a fault distribution map.

[0059] This embodiment discloses the collection of multi-source operating parameters of a photovoltaic (PV) panel, and the preprocessing of these parameters through an edge computing gateway to obtain preprocessed real-time operating parameters. The preprocessed real-time operating parameters are then fused with historical fault data, and the fusion result is input into a preset PV panel health assessment model to generate a health status assessment result. Based on the health status assessment result, a feature parsing algorithm is used to determine the fault type and fault location information of the PV panel. The fault type and fault location information are then spatiotemporally aligned to generate a fault distribution map of the PV panel. Because this embodiment preprocesses the multi-source operating parameters through an edge computing gateway to obtain preprocessed real-time operating parameters, and then fuses these real-time operating parameters with historical fault data before inputting them into a preset PV panel health assessment model, and then uses a feature parsing algorithm to determine the fault type and fault location information of the PV panel and generate a fault distribution map, this embodiment improves the accuracy and reliability of PV panel fault diagnosis compared to existing technologies.

[0060] refer to Figure 2 , Figure 2 This is a flowchart illustrating the second embodiment of the photovoltaic panel fault diagnosis method of the present invention.

[0061] Based on the first embodiment described above, in this embodiment, after step S40, steps S401 to S403 are further included: Step S401: Based on the fault category label corresponding to the fault type information, query the operation and maintenance strategy library for the handling strategy that matches the fault category label.

[0062] Step S402: Based on the handling strategy and the fault location information, generate operation and maintenance instructions and send the operation and maintenance instructions to the corresponding operation and maintenance terminal.

[0063] Step S403: After receiving the maintenance instruction, the maintenance terminal performs corresponding maintenance operations on the photovoltaic panel according to the maintenance instruction.

[0064] It should be understood that fault category labels can be used to uniquely identify a certain type of fault within the system. Labels can be integers, strings, or enumerated values ​​to allow for quick retrieval and matching by the system.

[0065] Furthermore, the operation and maintenance strategy library can refer to an offline database or knowledge graph that is continuously updated, storing the mapping relationship between fault category labels and handling strategies. Each record should at least include fault category labels, priority, suggested handling methods, required tools or spare parts, estimated working hours, and safety precautions, providing a rule source for automatically generating operation and maintenance instructions.

[0066] The handling strategy can be a specific work plan corresponding to the fault category label. For example, hot spot fault: immediately stop the machine and replace the diode or the entire component; dust accumulation fault or occlusion fault: call in a cluster of cleaning robots to clean; microcrack fault: mark the location, arrange for inspection and replacement.

[0067] It should be understood that maintenance instructions can refer to executable electronic work orders automatically generated by the system based on handling strategies and fault location information. Maintenance instructions are issued to the corresponding maintenance terminal and serve as the sole basis for on-site operations. Maintenance terminals can be receiving and feedback devices carried by on-site personnel or unmanned equipment, such as maintenance apps on mobile phones, or vehicle-mounted control boxes for drones or cleaning robots.

[0068] For example, when the fault type information corresponds to the fault category label of the photovoltaic panel having a dust accumulation fault or a shading fault, a cleaning instruction is generated based on the coordinates of the shading area in the fault distribution map; the cleaning instruction controls the adaptive cleaning robot cluster to plan the cleaning path of the photovoltaic panel through a reinforcement learning algorithm, and collects humidity sensor signals in real time during the path cleaning process; the cleaning path is adjusted based on the humidity sensor signals to obtain the adjusted cleaning path.

[0069] It should be noted that, in response to the problems of dust accumulation and obstruction, this embodiment uses an adaptive cleaning robot cluster to dynamically plan the cleaning path through reinforcement learning, and combines it with a humidity sensor to avoid secondary pollution.

[0070] This embodiment discloses a method for generating maintenance instructions based on fault category tags corresponding to the fault type information, querying a maintenance strategy library for handling strategies matching the fault category tags, generating maintenance instructions based on the handling strategies and the fault location information, and sending the maintenance instructions to the corresponding maintenance terminal. After receiving the maintenance instructions, the maintenance terminal performs corresponding maintenance operations on the photovoltaic panel according to the maintenance instructions. Compared with the prior art, this embodiment generates maintenance instructions based on handling strategies and fault location information, and sends the maintenance instructions to the corresponding maintenance terminal, realizing intelligent and precise scheduling of maintenance decisions, improving maintenance response efficiency and maintenance costs.

[0071] refer to Figure 3 , Figure 3 This is a flowchart illustrating the third embodiment of the photovoltaic panel fault diagnosis method of the present invention.

[0072] Based on the above embodiments, in this embodiment, after step S40, steps S404 to S406 are further included: Step S404: Input the historical fault records and the health status assessment results into the digital twin platform, and simulate the degradation process of each component in the photovoltaic panel through the digital twin platform to obtain the simulation results.

[0073] Step S405: Compare the simulation results with a preset degradation threshold to obtain a comparison result.

[0074] Step S406: When the comparison result meets the warning triggering conditions, generate warning information and push the warning information to the operation and maintenance management platform.

[0075] It should be noted that historical fault records can refer to all fault events that have occurred since the photovoltaic power station was put into operation, including at least: fault type, discovery time, location coordinates, environmental and electrical parameters at the time, handling method, replacement component model and performance recovery rate after repair; used to provide the digital twin platform with the initial conditions for component degradation and the basis for parameter calibration.

[0076] It should be understood that a digital twin platform can be a virtual mapping system built in the cloud or at the edge that runs synchronously with a real power plant. It can simulate material aging, power decay and potential failure evolution on a component-by-component basis and output a predicted value of the health status at any future time.

[0077] Understandably, by simulating the component degradation process through a digital twin platform, early warnings can be issued before a failure actually occurs, enabling a shift from reactive maintenance to predictive maintenance. Furthermore, after an early warning is issued, the information is pushed to the operation and maintenance management platform to promptly arrange precise repairs, thereby reducing power generation downtime caused by sudden failures and minimizing power generation losses.

[0078] Furthermore, by simulating the component degradation process through a digital twin platform, early warning information is pushed to the operation and maintenance management platform to arrange precise maintenance in a timely manner, avoiding large-scale inspections and excessive replacements, and saving operation and maintenance costs.

[0079] This embodiment discloses a method for inputting historical fault records and health status assessment results into a digital twin platform. The digital twin platform simulates the degradation process of each component in the photovoltaic panel to obtain simulation results. These simulation results are then compared with a preset degradation threshold to obtain comparison results. When the comparison results meet the early warning triggering conditions, an early warning message is generated and pushed to the operation and maintenance management platform. Compared to existing technologies, this embodiment, by simulating the component degradation process through a digital twin platform, enables the issuance of early warning information before the actual occurrence of a fault, reducing power generation loss and saving operation and maintenance costs.

[0080] Furthermore, this embodiment of the invention also proposes a storage medium storing a photovoltaic panel fault diagnosis program, which, when executed by a processor, implements the steps of the photovoltaic panel fault diagnosis method described above.

[0081] Reference Figure 4 , Figure 4This is a structural block diagram of the first embodiment of the photovoltaic panel fault diagnosis device of the present invention.

[0082] like Figure 4 As shown, the photovoltaic panel fault diagnosis device proposed in this embodiment of the invention includes: a parameter acquisition module 501, a status assessment module 502, a fault determination module 503, and an information processing module 504.

[0083] The parameter acquisition module 501 is used to acquire multi-source operating parameters of the photovoltaic panel and preprocess the multi-source operating parameters through the edge computing gateway to obtain preprocessed real-time operating parameters.

[0084] The status assessment module 502 is used to fuse the preprocessed real-time operating parameters and historical fault data, and input the fusion result into the preset photovoltaic panel health assessment model to generate a health status assessment result.

[0085] The fault determination module 503 is used to determine the fault type information and fault location information of the photovoltaic panel based on the health status assessment results and through a feature parsing algorithm.

[0086] The information processing module 504 is used to perform spatiotemporal alignment of the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel.

[0087] The status assessment module 502 is further configured to convert the preprocessed real-time operating parameters into time-series feature vectors, map the historical fault data into fault prior weights, and perform weighted fusion of the fault prior weights and the time-series feature vectors through a transfer learning mechanism to obtain a fusion result; and input the fusion result into a preset photovoltaic panel health assessment model to generate a health status assessment result.

[0088] The fault determination module 503 is further configured to use a feature parsing algorithm to parse the key features in the health status assessment result to obtain the parsing result; match the parsing result with a preset fault knowledge base, and determine the fault type information of the photovoltaic panel based on the matching result; trigger a distributed inspection task based on the fault type information, and receive the fault location information returned by the distributed inspection task.

[0089] The fault determination module 503 is further configured to create a distributed inspection task in the task scheduling queue according to the fault type information, and send the distributed inspection task to the corresponding UAV; acquire infrared-visible dual-mode images through the UAV, and input the dual-mode images into a lightweight generative network to obtain a defect mask image; bind the defect mask image with the BeiDou positioning data transmitted back by the UAV in real time to generate defect pixel blocks with geographic coordinates; and cluster and merge the defect pixel blocks to obtain fault location information.

[0090] This device embodiment discloses the acquisition of multi-source operating parameters of photovoltaic panels, and the preprocessing of these parameters through an edge computing gateway to obtain preprocessed real-time operating parameters. The preprocessed real-time operating parameters are then fused with historical fault data, and the fusion result is input into a preset photovoltaic panel health assessment model to generate a health status assessment result. Based on the health status assessment result, a feature parsing algorithm is used to determine the fault type information and fault location information of the photovoltaic panel. The fault type information and fault location information are then spatiotemporally aligned to generate a fault distribution map of the photovoltaic panel. Because this device embodiment preprocesses the multi-source operating parameters through an edge computing gateway to obtain preprocessed real-time operating parameters, and then fuses these real-time operating parameters with historical fault data before inputting them into a preset photovoltaic panel health assessment model, and finally uses a feature parsing algorithm to determine the fault type information and fault location information of the photovoltaic panel and generate a fault distribution map, compared to existing technologies, this device embodiment improves the accuracy and reliability of photovoltaic panel fault diagnosis.

[0091] Based on the first embodiment of the photovoltaic panel fault diagnosis device of the present invention, a second embodiment of the photovoltaic panel fault diagnosis device of the present invention is proposed.

[0092] In this embodiment, the information processing module 504 is further configured to query the operation and maintenance strategy library for a handling strategy that matches the fault category tag according to the fault type information; generate an operation and maintenance instruction based on the handling strategy and the fault location information, and send the operation and maintenance instruction to the corresponding operation and maintenance terminal; after the operation and maintenance terminal receives the operation and maintenance instruction, it performs corresponding operation and maintenance operations on the photovoltaic panel according to the operation and maintenance instruction.

[0093] The information processing module 504 is further configured to generate a cleaning instruction based on the coordinates of the shading area in the fault distribution map when the fault category label corresponding to the fault type information is a dust accumulation fault or a shading fault of the photovoltaic panel; control the adaptive cleaning robot cluster to plan the cleaning path of the photovoltaic panel through a reinforcement learning algorithm through the cleaning instruction, and collect humidity sensor signals in real time during the path cleaning process; adjust the cleaning path based on the humidity sensor signals to obtain the adjusted cleaning path.

[0094] Other embodiments or specific implementations of the photovoltaic panel fault diagnosis device of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0095] This application provides a photovoltaic panel fault diagnosis device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the photovoltaic panel fault diagnosis method in the above embodiment 1.

[0096] The following is for reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing a photovoltaic panel fault diagnosis device according to embodiments of this application. The photovoltaic panel fault diagnosis device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The photovoltaic panel fault diagnosis device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0097] like Figure 5As shown, the photovoltaic panel fault diagnosis device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the photovoltaic panel fault diagnosis device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the photovoltaic panel fault diagnosis equipment to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows photovoltaic panel fault diagnosis equipment with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0098] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0099] The photovoltaic panel fault diagnosis device provided in this application, employing the photovoltaic panel fault diagnosis method in the above embodiments, can solve the technical problem of low accuracy and reliability in existing photovoltaic panel fault diagnosis technologies. Compared with the prior art, the beneficial effects of the photovoltaic panel fault diagnosis device provided in this application are the same as those of the photovoltaic panel fault diagnosis method provided in the above embodiments, and other technical features of this photovoltaic panel fault diagnosis device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0100] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0102] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0103] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0104] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0105] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for diagnosing photovoltaic panel faults, characterized in that, The method includes: Multi-source operating parameters of photovoltaic panels are collected, and the multi-source operating parameters are preprocessed through an edge computing gateway to obtain preprocessed real-time operating parameters; The preprocessed real-time operating parameters and historical fault data are fused together, and the fusion result is input into a preset photovoltaic panel health assessment model to generate a health status assessment result. Based on the health status assessment results, the fault type information and fault location information of the photovoltaic panel are determined by feature parsing algorithm; The fault type information and the fault location information are spatiotemporally aligned to generate a fault distribution map of the photovoltaic panel; The step of determining the fault type information and fault location information of the photovoltaic panel based on the health status assessment results using a feature parsing algorithm includes: The key features in the health status assessment results are analyzed using a feature parsing algorithm to obtain the analysis results; The analysis results are matched with a preset fault knowledge base, and the fault type information of the photovoltaic panel is determined based on the matching results. Based on the fault type information, a distributed inspection task is created in the task scheduling queue, and the distributed inspection task is sent to the corresponding UAV. The UAV acquires infrared-visible dual-mode images, and the dual-mode images are input into a lightweight generative network to obtain defect mask images. The defect mask image is bound to the BeiDou positioning data transmitted back by the UAV in real time to generate defect pixel blocks with geographic coordinates. Clustering and merging the defective pixel blocks yields fault location information; After the step of spatiotemporally aligning the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel, the method further includes: Historical fault records and the health status assessment results are input into the digital twin platform, and the degradation process of each component in the photovoltaic panel is simulated through the digital twin platform to obtain simulation results; The simulation results are compared with a preset degradation threshold to obtain a comparison result; When the comparison result meets the warning triggering conditions, a warning message is generated and pushed to the operation and maintenance management platform; The step of fusing the preprocessed real-time operating parameters and historical fault data, and inputting the fusion result into a preset photovoltaic panel health assessment model to generate a health status assessment result includes: The preprocessed real-time operating parameters are converted into time-series feature vectors, and the historical fault data is mapped into fault prior weights. The prior fault weights and the temporal feature vectors are weighted and fused using a transfer learning mechanism to obtain a fusion result. The fusion results are input into a preset photovoltaic panel health assessment model to generate a health status assessment result.

2. The photovoltaic panel fault diagnosis method as described in claim 1, characterized in that, After the step of spatiotemporally aligning the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel, the method further includes: Based on the fault type information corresponding to the fault category label, query the operation and maintenance strategy library for the handling strategy that matches the fault category label; Based on the handling strategy and the fault location information, an operation and maintenance instruction is generated and sent to the corresponding operation and maintenance terminal. After receiving the maintenance instruction, the maintenance terminal performs corresponding maintenance operations on the photovoltaic panel according to the maintenance instruction.

3. The photovoltaic panel fault diagnosis method as described in claim 1, characterized in that, The method further includes: When the fault type information corresponds to the fault category label of the photovoltaic panel having a dust accumulation fault or a shading fault, a cleaning instruction is generated based on the coordinates of the shading area in the fault distribution map; The cleaning instructions control an adaptive cleaning robot cluster to plan the cleaning path of the photovoltaic panel using a reinforcement learning algorithm, and collect humidity sensor signals in real time during the cleaning process. The cleaning path is adjusted based on the humidity sensor signal to obtain the adjusted cleaning path.

4. A photovoltaic panel fault diagnosis device, characterized in that, The device includes: The parameter acquisition module is used to acquire multi-source operating parameters of the photovoltaic panel and preprocess the multi-source operating parameters through the edge computing gateway to obtain the preprocessed real-time operating parameters. The status assessment module is used to fuse the preprocessed real-time operating parameters and historical fault data, and input the fusion result into the preset photovoltaic panel health assessment model to generate a health status assessment result. The fault determination module is used to determine the fault type information and fault location information of the photovoltaic panel based on the health status assessment results and through a feature parsing algorithm. An information processing module is used to perform spatiotemporal alignment of the fault type information and the fault location information to generate a fault distribution map of the photovoltaic panel; The fault determination module is further configured to: analyze key features in the health status assessment results using a feature parsing algorithm to obtain analysis results; match the analysis results with a preset fault knowledge base and determine the fault type information of the photovoltaic panel based on the matching results; create a distributed inspection task in the task scheduling queue based on the fault type information and send the distributed inspection task to the corresponding UAV; acquire infrared-visible dual-mode images through the UAV and input the dual-mode images into a lightweight generative network to obtain a defect mask image; bind the defect mask image with the BeiDou positioning data transmitted back by the UAV in real time to generate defect pixel blocks with geographic coordinates; and cluster and merge the defect pixel blocks to obtain fault location information. The information processing module is also used to input historical fault records and the health status assessment results into the digital twin platform, simulate the degradation process of each component in the photovoltaic panel through the digital twin platform, and obtain simulation results; compare the simulation results with a preset degradation threshold to obtain comparison results; when the comparison results meet the early warning triggering conditions, generate early warning information and push the early warning information to the operation and maintenance management platform; The status assessment module is further configured to convert the preprocessed real-time operating parameters into time-series feature vectors, map the historical fault data into fault prior weights, and perform weighted fusion of the fault prior weights and the time-series feature vectors through a transfer learning mechanism to obtain a fusion result. The fusion result is then input into a preset photovoltaic panel health assessment model to generate a health status assessment result.

5. A photovoltaic panel fault diagnosis device, characterized in that, The device includes: a memory, a processor, and a photovoltaic panel fault diagnosis program stored in the memory and executable on the processor, the photovoltaic panel fault diagnosis program being configured to implement the steps of the photovoltaic panel fault diagnosis method as described in any one of claims 1 to 3.

6. A storage medium, characterized in that, The storage medium stores a photovoltaic panel fault diagnosis program, which, when executed by a processor, implements the steps of the photovoltaic panel fault diagnosis method as described in any one of claims 1 to 3.

Images