Real-time fault monitoring method and device for photovoltaic system, and medium

Abstract

The invention discloses a real-time fault monitoring method and device for a photovoltaic system and a medium, belongs to the technical field of photovoltaic system monitoring, and is used for solving the problems that real-time monitoring and fault monitoring of an existing photovoltaic power generation system have relatively high delay, cloud response delay is caused by a centralized data processing architecture, and fault monitoring is poor. And the fault early warning reaction time is relatively delayed. The method comprises the following steps: performing data acquisition on each photovoltaic module in a photovoltaic system through a distributed sensing array to obtain photovoltaic data; according to the 5G edge computing node, fault type analysis processing related to transmission data is carried out on the photovoltaic data, and fault data is determined; performing hierarchical transmission related to 5G network slices on the fault data, and determining priority transmission configuration of the fault data; performing health degree evaluation on the transmitted fault data through the priority transmission configuration to obtain a health degree evaluation report; and carrying out early warning calculation on the photovoltaic system under time-space association to obtain photovoltaic early warning information.

Description

Technical Field

[0001] This application relates to the field of photovoltaic system monitoring, and particularly to a real-time fault monitoring method, device and medium for a photovoltaic system. Background Art

[0002] In the field of photovoltaic power generation system monitoring technology, there are generally technical problems such as complex traditional wired transmission wiring, poor system scalability, and high fault response latency. Some related technologies have been disclosed in the prior art to solve these problems, but there are still many deficiencies.

[0003] In the existing monitoring of photovoltaic power generation systems, the real-time performance of the system is insufficient, resulting in a long fault confirmation time; second, there is a lack of effective monitoring and analysis of the dynamic changes of environmental parameters, and it is impossible to accurately judge compound faults; third, the data processing process is complex, which may lead to data transmission delays and low processing efficiency.

[0004] Moreover, the current monitoring of photovoltaic power generation systems generally has the following technical defects: traditional wired transmission wiring is complex, the system scalability is poor, and new devices need to be rewired. The wireless transmission method has problems such as high latency (typical value > 50ms) and limited bandwidth. The centralized data processing architecture leads to cloud response delays, and the fault confirmation time exceeds 10 minutes. The parameter acquisition dimension is single (only voltage / current), and it is impossible to accurately judge compound faults. At the same time, the existing fault diagnosis algorithms do not consider the dynamic coupling effect of environmental parameters (such as the influence of temperature on the IV curve), and the fault warning response time is relatively delayed. Summary of the Invention

[0005] Embodiments of this application provide a real-time fault monitoring method, device and medium for a photovoltaic system to solve the following technical problems: there is a high delay in the real-time monitoring and fault monitoring of existing photovoltaic power generation systems, the centralized data processing architecture leads to cloud response delays, and the fault warning response time is relatively delayed.

[0006] Embodiments of this application adopt the following technical solutions:

[0007] On the one hand, embodiments of this application provide a real-time fault monitoring method for a photovoltaic system, including: collecting data of each photovoltaic component in the photovoltaic system through a distributed sensing array to obtain photovoltaic data; analyzing and processing the fault types of the transmission data for the photovoltaic data according to a 5G edge computing node to determine fault data; performing hierarchical transmission of the fault data related to a 5G network slice to determine the priority transmission configuration of the fault data; evaluating the health of the transmitted fault data through the priority transmission configuration to obtain a health evaluation report; and performing early warning calculation under spatio-temporal association on the photovoltaic system according to the health evaluation report to obtain photovoltaic early warning information.

[0008] In the embodiment of the present application, data is collected for each photovoltaic component in the photovoltaic system through a distributed sensing array, realizing real-time monitoring of the state of the photovoltaic system, which helps to promptly discover potential problems. Using a 5G edge computing node to analyze and process the photovoltaic data can quickly determine the faulty data, improving the efficiency of fault diagnosis. By analyzing and processing to determine the faulty data, it helps to accurately locate the position where the fault occurs, reducing misjudgment and misoperation of the entire system. Hierarchical transmission of the faulty data through 5G network slicing can allocate network resources according to the urgency of the fault, ensuring that critical fault information can be preferentially transmitted. Through the priority transmission configuration, it can ensure that the emergency fault data is processed in a timely manner, reducing the impact of the fault on the operation of the photovoltaic system. Conducting a health assessment on the transmitted faulty data can comprehensively understand the health status of the photovoltaic system, providing a basis for maintenance and optimization.

[0009] In a feasible implementation manner, data is collected for each photovoltaic component in the photovoltaic system through a distributed sensing array to obtain photovoltaic data, specifically including: deploying the distributed sensing array into each photovoltaic component; wherein, the distributed sensing array includes: an IV curve dynamic scanning unit, an infrared temperature sensor, and a light intensity monitoring module; through the programmable load array and the high-speed data acquisition card in the IV curve dynamic scanning unit, full-range scanning of the voltage and current in each photovoltaic component is performed to obtain the IV curve dynamic data at the current moment; through the infrared temperature sensor, the temperature data of each photovoltaic component at the current moment is collected; based on the light intensity monitoring module, the working power data of each photovoltaic component at the current moment is recorded; the IV curve dynamic data, the temperature data, and the working power data are fused to obtain the photovoltaic data of each photovoltaic component at the current moment.

[0010] In a feasible implementation manner, according to the edge computing node, fault type analysis processing of the transmission data is performed on the photovoltaic data to determine fault data, which specifically includes: extracting physical features in the photovoltaic data; wherein, the physical features include: IV curve dynamic change features, power data change features, and temperature data change features; through a preset 1D-CNN network, deep extraction processing is performed on the time series data related to voltage and current in the IV curve dynamic change features, and based on the spatial features of each photovoltaic module time series data, IV spatio-temporal features are obtained; through an LSTM network, the long-term relationship between the power data change features and the light intensity data is captured to obtain power-light intensity dependence features; and the IV spatio-temporal features and the power-light intensity dependence features are combined to obtain the deep features of the photovoltaic data; based on the attention mechanism, multi-modal feature fusion processing is performed on the deep features and the physical features to obtain the index quantization features of the photovoltaic data; wherein, the index quantization features include multiple quantifiable index features in the photovoltaic data; through an incremental model updated by a 5G edge computing node and based on the federated learning of the cloud to update the global model, fault type classification processing of the index quantization features under a threshold interval is performed to obtain the fault data.

[0011] In a feasible implementation manner, the 5G edge computing node includes: a lightweight CNN model, which is used to compress the parameter quantity of the incremental model to less than 1MB by using the channel pruning technology; a local cache queue, which is used to store the original photovoltaic data in the recent 1 hour and support rolling update; an abnormal data interception module, which filters out abnormal values beyond the normal distribution based on the 3σ criterion.

[0012] In a feasible implementation manner, fault type classification processing of the index quantization features under a threshold interval is performed to obtain the fault data, which specifically includes: if the comprehensive fault score of each index feature in the index quantization features is greater than the second preset threshold, the fault type corresponding to the index quantization features is determined as an emergency alarm fault; if the comprehensive fault score of each index feature in the index quantization features is less than or equal to the second preset threshold and greater than the first preset threshold, the fault type corresponding to the index quantization features is determined as a conventional monitoring error fault; if the comprehensive fault score of each index feature in the index quantization features is less than or equal to the first preset threshold and the photovoltaic system is in a software upgrade state, the fault type corresponding to the index quantization features is determined as a software upgrade waiting fault; wherein, the fault data includes: emergency alarm fault, conventional monitoring error fault, and software upgrade waiting fault.

[0013] In a feasible implementation, the fault data is hierarchically transmitted regarding the 5G network slice, and the priority transmission configuration of the fault data is determined, which specifically includes: If the fault data is an emergency alarm fault, the data transmission channel of the 5G network slice is configured as an emergency alarm channel; wherein, the emergency alarm channel reserves 20% bandwidth and is the highest transmission priority channel; If the fault data is a conventional monitoring error fault, the data transmission channel of the 5G network slice is configured as a conventional monitoring channel; wherein, the conventional monitoring channel adopts a dynamic bandwidth allocation strategy and the allocation ratio is automatically adjusted according to the photovoltaic system load; If the fault data is a software upgrade waiting fault, the data transmission channel of the 5G network slice is configured as a software upgrade channel; wherein, the software upgrade channel only performs batch firmware updates during fixed periods.

[0014] In a feasible implementation, through the priority transmission configuration, the transmitted fault data is subjected to a health assessment to obtain a health assessment report, which specifically includes: Based on multiple health assessment indicators of each photovoltaic module, the health weight of each health assessment indicator is determined; All the transmitted fault data is counted; And based on the impact degree of the fault data, the fault score of each type of fault data is determined; According to the fault score of each type of fault data, a decentralization calculation of the health penalty factor for each type of fault data is performed to obtain the penalty weight; Through the health weight and the penalty weight, a health assessment of the current photovoltaic system under overall normal operation is performed, and the health assessment report is generated.

[0015] In a feasible implementation, according to the health assessment report, an early warning calculation of the photovoltaic system under spatio-temporal association is performed to obtain photovoltaic early warning information, which specifically includes: Through the cloud intelligent analysis platform, a threshold comparison is made on the health assessment score of the photovoltaic system at the current moment in the health assessment report to obtain the early warning information based on the time scale; The position area of the photovoltaic modules corresponding to the fault data in the health assessment report is divided to obtain the fault photovoltaic area; The early warning information and the fault photovoltaic area are integrated under spatio-temporal association to obtain the photovoltaic early warning information of the photovoltaic system.

[0016] In a second aspect, an embodiment of the present application further provides a real-time fault monitoring device for a photovoltaic system, the device includes: At least one processor; And a memory communicatively connected to the at least one processor; wherein, the memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute a real-time fault monitoring method for a photovoltaic system according to any one of the above embodiments.

[0017] In a third aspect, an embodiment of the present application further provides a non-volatile computer storage medium, characterized in that the storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, and each program includes instructions that, when executed by a terminal, cause the terminal to execute a real-time fault monitoring method for a photovoltaic system according to any one of the above embodiments.

[0018] The present application provides a real-time fault monitoring method, device, and medium for a photovoltaic system. Compared with the prior art, the embodiments of the present application have the following beneficial technical effects:

[0019] 1. Real-time monitoring: Data collection is performed on each photovoltaic module in the photovoltaic system through a distributed sensing array, realizing real-time monitoring of the state of the photovoltaic system, which helps to detect potential problems in a timely manner.

[0020] 2. Efficient fault analysis: Using 5G edge computing nodes to analyze and process the photovoltaic data can quickly determine the fault data, improving the efficiency of fault diagnosis.

[0021] 3. Precise fault location: By analyzing and processing to determine the fault data, it helps to accurately locate the position where the fault occurs, reducing misjudgment and misoperation of the entire system.

[0022] 4. Hierarchical transmission: Hierarchical transmission of the fault data through 5G network slicing can allocate network resources according to the urgency of the fault, ensuring that critical fault information can be transmitted preferentially.

[0023] 5. Preferential transmission configuration: Through the preferential transmission configuration, it can ensure that emergency fault data is processed in a timely manner, reducing the impact of faults on the operation of the photovoltaic system.

[0024] 6. Health assessment: Conducting a health assessment on the transmitted fault data can comprehensively understand the health status of the photovoltaic system, providing a basis for maintenance and optimization.

[0025] 7. Early warning calculation: Based on the health assessment report, early warning calculation under spatio-temporal association can predict potential risks in advance, take preventive measures, and avoid the occurrence of faults.

[0026] 8. Improve system reliability: Through real-time monitoring and early warning, the reliability of the photovoltaic system can be significantly improved, reducing the downtime caused by faults.

[0027] 9. Reduce maintenance costs: Through early warning and preventive maintenance, the maintenance costs at the time of fault occurrence can be reduced, and the service life of the system can be extended.

[0028] 10. Improve operation and maintenance efficiency: The integrated monitoring, analysis, and early warning system can improve the efficiency of operation and maintenance personnel and reduce the workload of manual inspections. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments recorded in the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings:

[0030] Figure 1 It is a flowchart of a real-time fault monitoring method for a photovoltaic system provided by an embodiment of the present application;

[0031] Figure 2 It is a schematic structural diagram of a real-time fault monitoring device for a photovoltaic system provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to enable those skilled in the art to better understand the technical solutions in the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0033] An embodiment of the present application provides a real-time fault monitoring method for a photovoltaic system. As Figure 1 shown, the real-time fault monitoring method for a photovoltaic system specifically includes steps S101 - S105:

[0034] S101. Through a distributed sensing array, data is collected for each photovoltaic module in the photovoltaic system to obtain photovoltaic data.

[0035] Specifically, it is necessary to first deploy the distributed sensing array into each photovoltaic module. Among them, the distributed sensing array includes: an IV curve dynamic scanning unit, an infrared temperature sensor, and a light intensity monitoring module.

[0036] Further, through the programmable load array and high-speed data acquisition card in the IV curve dynamic scanning unit, a full-range scan of the voltage and current in each photovoltaic module is performed to obtain the IV curve dynamic data at the current moment.

[0037] Further, through the infrared temperature sensor, the temperature data of each photovoltaic module at the current moment is collected.

[0038] Further, based on the light intensity monitoring module, record the working power data of each photovoltaic module at the current moment.

[0039] Further, perform data fusion on the IV curve dynamic data, temperature data, and working power data to obtain the photovoltaic data of each photovoltaic module at the current moment.

[0040] In one embodiment, select a suitable distributed sensing array, including an IV curve dynamic scanning unit, an infrared temperature sensor, and a light intensity monitoring module. Clean and inspect each photovoltaic module to ensure that the surface is free of dust and dirt to obtain accurate measurement data. Fix the IV curve dynamic scanning unit on one side of the photovoltaic module to ensure that it can irradiate sunlight without obstruction. Paste or fix the infrared temperature sensor on the back or side of the photovoltaic module to avoid direct exposure to sunlight to accurately measure the component temperature. Install the light intensity monitoring module near the photovoltaic module to ensure that the light intensity received by the module can be accurately measured. After completing the installation of all sensors, conduct a preliminary test to ensure that each sensor can work properly.

[0041] In one embodiment, collect the data of the photovoltaic module through the IV curve dynamic scanning unit, the infrared temperature sensor, and the light intensity monitoring module. Use the programmable load array in the IV curve dynamic scanning unit to perform a full-range scan on each photovoltaic module. Real-time record the voltage and current data through a high-speed data acquisition card. Analyze the scanned data to generate the IV curve dynamic data at the current moment. Collect the temperature data of each photovoltaic module at the current moment through the infrared temperature sensor. Based on the light intensity monitoring module, record the working power data of each photovoltaic module at the current moment. Perform fusion on the collected IV curve dynamic data, temperature data, and working power data. Through a data fusion algorithm, combine different types of data to obtain the comprehensive photovoltaic data of each photovoltaic module at the current moment. Process and analyze the collected photovoltaic data to evaluate the performance and health status of the photovoltaic module. Use appropriate software tools to process the fused data, including data cleaning, outlier detection, and trend analysis. Compare the processed data with historical data to identify any signs of performance degradation or potential failures. Generate a performance report and a health assessment of the photovoltaic module according to the analysis results.

[0042] S102. According to the 5G edge computing node, perform analysis and processing on the photovoltaic data for the fault types of transmitted data to determine the fault data.

[0043] Specifically, first extract the physical features in the photovoltaic data. Among them, the physical features include: the dynamic change features of the IV curve, the change features of the power data, and the change features of the temperature data.

[0044] Furthermore, through a preset 1D-CNN network, the time-series data related to voltage and current in the dynamic change characteristics of the IV curve is deeply extracted, and based on the spatial characteristics of the time-series data of each photovoltaic module, the IV spatio-temporal characteristics are obtained.

[0045] Furthermore, it is also necessary to capture the long-term relationship between the change characteristics of power data and light intensity data through an LSTM network to obtain the power-light intensity dependence characteristics. Then, the IV spatio-temporal characteristics and the power-light intensity dependence characteristics are combined to obtain the deep characteristics of photovoltaic data.

[0046] Furthermore, based on the attention mechanism, the deep characteristics and physical characteristics are subjected to multi-modal feature fusion processing to obtain the index quantization characteristics of photovoltaic data. Among them, the index quantization characteristics include multiple quantifiable index characteristics in the photovoltaic data.

[0047] Furthermore, through the incremental model updated by the 5G edge computing node and based on the federated learning of the cloud to update the global model, the fault type classification processing of the index quantization characteristics under the threshold interval is carried out to obtain the fault data.

[0048] As a feasible implementation, the 5G edge computing node includes: a lightweight CNN model, which is used to compress the parameter quantity of the incremental model to less than 1MB by using the channel pruning technology; a local cache queue, which is used to store the original photovoltaic data in the recent 1 hour and support rolling update; and an abnormal data interception module, which filters out the abnormal values beyond the normal distribution based on the 3σ criterion.

[0049] As a feasible implementation, if the comprehensive fault score of each index characteristic in the index quantization characteristics is greater than the second preset threshold, the fault type corresponding to the index quantization characteristics is determined as an emergency warning fault. If the comprehensive fault score of each index characteristic in the index quantization characteristics is less than or equal to the second preset threshold and greater than the first preset threshold, the fault type corresponding to the index quantization characteristics is determined as a conventional monitoring error fault. If the comprehensive fault score of each index characteristic in the index quantization characteristics is less than or equal to the first preset threshold and the photovoltaic system is in the software upgrade state, the fault type corresponding to the index quantization characteristics is determined as a software upgrade waiting fault. Among them, the fault data includes: emergency warning faults, conventional monitoring error faults, and software upgrade waiting faults.

[0050] In one embodiment, a three - level architecture including a photovoltaic component monitoring device, a 5G edge computing node, and a cloud server is constructed: Monitoring layer: Each photovoltaic component is equipped with an IV curve sensor (sampling rate 1Hz), a power sensor (±0.5% accuracy), and an infrared temperature probe (range - 20°C to 150°C). Edge layer: A 5G MEC node (computing power ≥8 TOPS) deployed in the photovoltaic power station, with an incremental LSTM - CNN hybrid model built in. Cloud layer: Adopt a Kubernetes cluster to manage the federated learning framework and aggregate global model parameters.

[0051] In one embodiment, the dynamic features of the IV curve: 10 key points (V_oc, I_sc, V_mpp, I_mpp, etc.) are extracted through discretization processing to construct a time - series matrix X_iv∈R^{N×10×2} (N is the time step). Power feature: Calculate the first - order derivative of ΔP / Δt as the dynamic change index, and the window length is set to 5 minutes. Temperature feature: Use spatial interpolation method to generate a temperature distribution heat map with a resolution of 0.1°C / pixel. In the implementation of the LSTM network (power - light intensity dependence modeling), input features: historical power sequence (length 60 minutes) and light intensity (from the meteorological station API). Network structure: Bidirectional LSTM (hidden_size = 64)+time attention mechanism.

[0052] In one embodiment, a cascaded attention mechanism is used to achieve feature fusion: Feature alignment: Map the CNN output (dimension 128) and the LSTM output (dimension 64) to a unified space (d_model = 256). Then perform cross - modal attention calculation: where Q = physical feature, K = V = deep feature. In incremental update: The edge node performs local training every 2 hours (learning rate 0.001, SGD optimizer). Federated aggregation: Adopt the FedProx algorithm and set μ = 0.1 to prevent model drift. Finally, construct a three - level threshold classifier: First - level classification (voltage anomaly): Threshold interval [V_nom×0.8, V_nom×1.2]. Second - level classification (temperature - power mismatch): Establish an envelope of the P - T curve, and an anomaly is determined if the deviation > 15%. Third - level classification (IV curve distortion): Use the DTW algorithm to calculate the distance from the standard curve, and the threshold d = 0.35.

[0053] S103. Perform hierarchical transmission of the fault data related to the 5G network slice to determine the priority transmission configuration of the fault data.

[0054] Specifically, if the fault data is an emergency alarm fault, configure the data transmission channel of the 5G network slice as an emergency alarm channel. Among them, the emergency alarm channel reserves 20% of the bandwidth and is the highest - priority transmission channel.

[0055] If the fault data is a conventional monitoring error fault, configure the data transmission channel of the 5G network slice as a conventional monitoring channel. Among them, the conventional monitoring channel adopts a dynamic bandwidth allocation strategy, and the allocation ratio is automatically adjusted according to the load of the photovoltaic system. If the fault data is a software upgrade waiting fault, configure the data transmission channel of the 5G network slice as a software upgrade channel. Among them, the software upgrade channel only performs batch firmware updates during fixed periods.

[0056] In one implementation, a distributed sensing array is deployed on each photovoltaic module to collect data such as voltage, current, temperature, and light intensity in real time. When a fault is detected, the system transmits the fault data to the 5G base station. The 5G base station performs a preliminary analysis of the fault data through an edge computing node and classifies it as an emergency warning fault, a conventional monitoring error fault, or a software upgrade waiting fault. If the fault data is an emergency warning fault, the system preferentially transmits the fault data to the emergency warning channel. The emergency warning channel is pre-configured with 20% of the bandwidth and set to the highest transmission priority. The 5G core network ensures that the emergency warning data is preferentially transmitted within 20% of the bandwidth and reaches the monitoring platform at the fastest speed. If the fault data is a conventional monitoring error fault, the system transmits the fault data to the conventional monitoring channel. The conventional monitoring channel adopts a dynamic bandwidth allocation strategy and automatically adjusts the bandwidth allocation ratio according to the load of the photovoltaic system. The monitoring platform monitors the load of the photovoltaic system in real time and dynamically adjusts the bandwidth allocation of the conventional monitoring channel. If the fault data is a software upgrade waiting fault, the system transmits the fault data to the software upgrade channel. The software upgrade channel only performs batch firmware updates during fixed maintenance periods. During the fixed period, the system will automatically centrally process the firmware update tasks of all photovoltaic modules to ensure the stable operation of the system.

[0057] As a feasible implementation, emergency warning faults are quickly responded to, reducing the impact of faults on the photovoltaic system. Conventional monitoring error faults optimize network resource utilization through dynamic bandwidth allocation. Software upgrade waiting faults are processed during fixed periods without affecting the normal operation of the photovoltaic system.

[0058] S104. Through the priority transmission configuration, evaluate the health status of the transmitted fault data to obtain a health status evaluation report.

[0059] Specifically, use multiple health status evaluation indicators of each photovoltaic module to determine the health weights of each health status evaluation indicator.

[0060] Furthermore, count all the transmitted fault data. And based on the impact degree of the fault data, determine the fault scores of each type of fault data.

[0061] Furthermore, according to the fault scores of the fault data, perform a decentralized calculation of the health penalty factors for the fault data to obtain the penalty weights.

[0062] Further, through the health weight and the penalty weight, the current photovoltaic system is evaluated for its health under overall normal operation, and a health evaluation report is generated.

[0063] S105. According to the health evaluation report, perform early warning calculation on the photovoltaic system under spatio-temporal correlation to obtain photovoltaic early warning information.

[0064] Specifically, in combination with the cloud intelligent analysis platform, compare the health evaluation score of the photovoltaic system at the current moment in the health evaluation report with a threshold value to obtain early warning information based on the time scale.

[0065] Further, divide the range of the position area of the photovoltaic modules corresponding to the fault data in the health evaluation report to obtain the faulty photovoltaic area.

[0066] Further, integrate the early warning information and the faulty photovoltaic area under spatio-temporal correlation to obtain the photovoltaic early warning information of the photovoltaic system.

[0067] In one embodiment, the cloud intelligent analysis platform uses machine learning algorithms to analyze the collected data and generate a health evaluation report. The report includes the health evaluation score of the photovoltaic system at the current moment. Set the threshold values of the health evaluation score, which are based on the historical data and expected performance of the photovoltaic system. Compare the current moment health evaluation score in the health evaluation report with the preset threshold values. If the current health evaluation score is lower than a certain threshold value, generate early warning information based on the time scale. Determine the faulty photovoltaic modules according to the fault data in the health evaluation report. Use geographic information system (GIS) technology to divide the range of the positions of the faulty photovoltaic modules and identify the faulty photovoltaic area.

[0068] In one embodiment, correlate the generated early warning information with the faulty photovoltaic area in space and time. Integrate data such as the position information, fault type, and occurrence time of the faulty photovoltaic area. Combine the early warning information and the faulty photovoltaic area data to generate the photovoltaic early warning information of the photovoltaic system. The early warning information may include fault descriptions, expected impacts, recommended maintenance measures, etc. The early warning information is classified, stored, and distributed through the early warning information management system. The user interface allows system administrators and operation and maintenance personnel to view, review, and respond to the early warning information.

[0069] In addition, the embodiment of the present application also provides a real-time fault monitoring device for a photovoltaic system, as Figure 2 shown, the real-time fault monitoring device 200 of the photovoltaic system specifically includes:

[0070] At least one processor 201. And, a memory 202 communicatively connected to the at least one processor 201. Wherein, the memory 202 stores instructions executable by the at least one processor 201, enabling the at least one processor 201 to execute:

[0071] Collect data of each photovoltaic component in the photovoltaic system through a distributed sensing array to obtain photovoltaic data.

[0072] Analyze and process the photovoltaic data for fault types related to transmission data according to a 5G edge computing node to determine fault data.

[0073] Perform hierarchical transmission of the fault data related to 5G network slicing to determine the priority transmission configuration of the fault data.

[0074] Evaluate the health of the transmitted fault data through the priority transmission configuration to obtain a health evaluation report.

[0075] Perform early warning calculation for the photovoltaic system under spatio-temporal association according to the health evaluation report to obtain photovoltaic early warning information.

[0076] In the embodiments of the present application, data of each photovoltaic component in the photovoltaic system is collected through a distributed sensing array, realizing real-time monitoring of the state of the photovoltaic system, which helps to detect potential problems in a timely manner. Using a 5G edge computing node to analyze and process the photovoltaic data for fault types can quickly determine the fault data, improving the efficiency of fault diagnosis. By analyzing and processing to determine the fault data, it helps to accurately locate the position where the fault occurs, reducing misjudgment and misoperation of the entire system. Performing hierarchical transmission of the fault data for 5G network slicing can allocate network resources according to the urgency of the fault, ensuring that critical fault information can be transmitted preferentially. Through the priority transmission configuration, it can ensure that emergency fault data is processed in a timely manner, reducing the impact of the fault on the operation of the photovoltaic system. Evaluating the health of the transmitted fault data can comprehensively understand the health status of the photovoltaic system, providing a basis for maintenance and optimization.

[0077] The various embodiments in the present application are all described in a progressive manner. The same or similar parts among the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the device and the non-volatile computer storage medium, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments.

[0078] The devices, media, and methods provided by the embodiments of this application are in one-to-one correspondence. Therefore, the devices and media also have beneficial technical effects similar to those of their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be elaborated here.

[0079] Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0080] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of this application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the flows or multiple flows and / or blocks Figure 1 one or more of the blocks or multiple blocks.

[0081] Memory may include non-permanent memory in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0082] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transitory medium that can store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0083] The above describes specific embodiments of the present application. In some cases, the recited actions or steps may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order shown or sequential order to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0084] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the embodiments of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the embodiments of the present application should be included in the scope of the present application.

Claims

1. A real-time fault monitoring method for a photovoltaic system, characterized in that: The method comprises: Through the distributed sensor array, data is collected from each photovoltaic module in the photovoltaic system to obtain photovoltaic data; According to the 5G edge computing node, the photovoltaic data is analyzed and processed for the fault type of the transmission data to determine the fault data; Transmitting the fault data in a hierarchical manner on the 5G network slice, and determining a priority transmission configuration for the fault data; Through the priority transmission configuration, the transmitted fault data is subjected to health assessment to obtain a health assessment report; According to the health assessment report, a warning calculation is performed on the photovoltaic system under time and space correlation to obtain photovoltaic warning information.

2. A real-time fault monitoring method for a photovoltaic system according to claim 1, characterized in that: Through the distributed sensor array, data is collected from each photovoltaic module in the photovoltaic system to obtain photovoltaic data, including: Deploy the distributed sensor array in each photovoltaic module; wherein the distributed sensor array includes: an IV curve dynamic scanning unit, an infrared temperature sensor and a light intensity monitoring module; Through the programmable load array and high-speed data acquisition card in the IV curve dynamic scanning unit, the voltage and current in each photovoltaic module are fully scanned to obtain the IV curve dynamic data at the current moment; The infrared temperature sensor is used to collect the temperature data of each photovoltaic module at the current moment; Based on the light intensity monitoring module, record the working power data of each photovoltaic component at the current moment; The IV curve dynamic data, the temperature data and the operating power data are fused to obtain photovoltaic data of each photovoltaic component at the current moment.

3. A real-time fault monitoring method for a photovoltaic system according to claim 1, characterized in that: According to the edge computing node, the photovoltaic data is analyzed and processed for the fault type of the transmission data to determine the fault data, specifically including: Extracting physical features from the photovoltaic data; wherein the physical features include: dynamic change features of the IV curve, power data change features, and temperature data change features; Through the preset 1D-CNN network, the time series data of voltage and current in the dynamic change characteristics of the IV curve are deeply extracted and processed, and the IV spatiotemporal characteristics are obtained based on the spatial characteristics of the time series data of each photovoltaic module; Through the LSTM network, the long-term relationship between the power data change characteristics and the light intensity data is captured to obtain the power light intensity dependence characteristics; and the IV spatiotemporal characteristics are combined with the power light intensity dependence characteristics to obtain the deep characteristics of the photovoltaic data; Based on the attention mechanism, the deep features and the physical features are subjected to multimodal feature fusion processing to obtain the index quantification features of the photovoltaic data; wherein the index quantification features include multiple quantifiable index features in the photovoltaic data; The incremental model after the 5G edge computing node is updated, and the global model is updated based on the federated learning in the cloud, and the fault type classification processing is performed on the quantitative characteristics of the indicator within the threshold range to obtain the fault data.

4. A real-time fault monitoring method for a photovoltaic system according to claim 3, characterized in that: The 5G edge computing node includes: A lightweight CNN model, wherein the lightweight CNN model is used to compress the incremental model parameters to less than 1MB by using a channel pruning technique; A local cache queue, which is used to store the original photovoltaic data of the last hour and supports rolling updates; The abnormal data interception module filters out abnormal values ​​beyond the normal distribution based on the 3σ criterion.

5. A real-time fault monitoring method for a photovoltaic system according to claim 3, characterized in that: The quantitative characteristics of the indicators are subjected to fault type classification processing within a threshold range to obtain the fault data, specifically including: If the comprehensive fault score of each indicator feature in the indicator quantitative feature is greater than the second preset threshold, the fault type corresponding to the indicator quantitative feature is determined as an emergency alarm fault; If the comprehensive fault score of each indicator feature in the indicator quantitative feature is less than or equal to the second preset threshold and greater than the first preset threshold, the fault type corresponding to the indicator quantitative feature is determined as a conventional monitoring error fault; If the comprehensive fault score of each indicator feature in the indicator quantification feature is less than or equal to the first preset threshold, and the photovoltaic system is in a software upgrade state, the fault type corresponding to the indicator quantification feature is determined as a software upgrade waiting fault; The fault data includes: emergency alarm fault, conventional monitoring error fault and software upgrade waiting fault.

6. A real-time fault monitoring method for a photovoltaic system according to claim 1, characterized in that: The fault data is hierarchically transmitted on the 5G network slice, and a priority transmission configuration of the fault data is determined, specifically including: If the fault data is an emergency alarm fault, the data transmission channel of the 5G network slice is configured as an emergency alarm channel; wherein the emergency alarm channel reserves 20% bandwidth and is the highest transmission priority channel; If the fault data is a conventional monitoring error fault, the data transmission channel of the 5G network slice is configured as a conventional monitoring channel; wherein the conventional monitoring channel adopts a dynamic bandwidth allocation strategy and the allocation ratio of the photovoltaic system load is automatically adjusted; If the fault data is a software upgrade waiting fault, the data transmission channel of the 5G network slice is configured as a software upgrade channel; wherein, the software upgrade channel performs batch firmware updates only during a fixed period of time.

7. A real-time fault monitoring method for a photovoltaic system according to claim 1, characterized in that: Through the priority transmission configuration, the transmitted fault data is subjected to health evaluation to obtain a health evaluation report, which specifically includes: Based on multiple health evaluation indicators of each photovoltaic component, determine the health weight of each health evaluation indicator; Counting all the fault data after transmission; and determining the fault score of each fault data based on the impact degree of the fault data; According to the fault score of each type of fault data, a weighted calculation of a health penalty factor is performed on each type of fault data to obtain a penalty weight; The health weight and the penalty weight are used to evaluate the health of the current photovoltaic system under overall normal operation, and the health evaluation report is generated.

8. A real-time fault monitoring method for a photovoltaic system according to claim 1, characterized in that: According to the health assessment report, the photovoltaic system is subjected to a warning calculation under time-space correlation to obtain photovoltaic warning information, which specifically includes: Through the cloud intelligent analysis platform, the health assessment score of the photovoltaic system at the current moment in the health assessment report is compared with the threshold value to obtain early warning information based on the time scale; Divide the photovoltaic components corresponding to the fault data in the health assessment report into a range of location areas to obtain a faulty photovoltaic area; The warning information is integrated with the faulty photovoltaic area in time and space correlation to obtain the photovoltaic warning information of the photovoltaic system.

9. A real-time fault monitoring device for a photovoltaic system, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the real-time fault monitoring method for a photovoltaic system according to any one of claims 1-8.

10. A non-volatile computer storage medium, characterized in that: The storage medium is a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium stores at least one program, each of which includes instructions, and when the instructions are executed by the terminal, the terminal executes a real-time fault monitoring method for a photovoltaic system according to any one of claims 1-8.

Images