Photovoltaic inverter test data analysis system and method based on edge calculation

Abstract

The invention discloses a photovoltaic inverter test data analysis system and method based on edge calculation, and relates to the related field of data processing, and the system comprises a data collection and fusion layer which is used for obtaining electrical parameters, environment data and thermal imaging data, and carrying out the fusion preprocessing of the data; the edge intelligent cooperation layer is used for constructing an intelligent cooperation network, including internal cooperation communication and external federated privacy communication, and updating a prediction test channel of the edge intelligent gateway; the adaptive optimization layer is used for receiving real-time response parameters of the photovoltaic inverter and synchronously sending a fusion preprocessing result to the prediction test channel for adaptive analysis; and the correction layer is used for receiving the adaptive analysis result and executing control optimization of the photovoltaic inverter. The technical problem that the real-time performance and privacy protection capability of data processing are insufficient in existing photovoltaic inverter test data analysis is solved, and the technical effect of improving the real-time performance and privacy of data processing is achieved.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to a photovoltaic inverter test data analysis system and method based on edge computing. Background Technology

[0002] As the core equipment in a photovoltaic (PV) power generation system, the photovoltaic (PV) inverter is responsible for converting the direct current (DC) generated by PV modules into alternating current (AC) to supply the grid or loads. The performance, stability, and efficiency of the PV inverter directly affect the power generation and operating costs of the entire PV power generation system. Therefore, efficient and accurate testing and data analysis of PV inverters are crucial for improving the overall performance of PV power generation systems. Existing PV inverter testing and data analysis systems primarily rely on a centralized data processing architecture. This architecture involves a large server cluster deployed in a remote data center, responsible for collecting data from various PV inverters and performing unified processing and analysis. This method suffers from data latency due to long-distance transmission to the remote data center, affecting real-time performance. Furthermore, the transmission of large amounts of sensitive data over public networks increases the risk of data leakage.

[0003] Currently, photovoltaic inverter test data analysis suffers from technical issues such as insufficient real-time data processing and inadequate privacy protection capabilities. Summary of the Invention

[0004] This application provides a photovoltaic inverter test data analysis system and method based on edge computing. It employs methods such as directly collecting electrical parameters, environmental data, and thermal imaging data near the photovoltaic inverter, performing data fusion and preprocessing through an edge smart gateway to reduce data transmission volume, constructing an intelligent collaborative network, and enabling internal collaborative communication between edge smart gateways and external federated privacy communication. This ensures data sharing and collaboration while protecting privacy. The system dynamically updates the predictive test channel based on real-time data, receives real-time response parameters from the photovoltaic inverter based on the predictive test channel, and simultaneously sends the fusion preprocessing results. It performs adaptation analysis of the real-time response parameters and, based on the adaptation analysis results, executes control optimization of the photovoltaic inverter to ensure that the photovoltaic inverter always operates in its optimal state. By pushing data processing capabilities down to the vicinity of the data source, i.e., the edge node where the photovoltaic inverter is located, it achieves real-time data acquisition, preprocessing, and analysis, significantly improving the real-time performance of data processing. Combined with encryption technology and privacy protection mechanisms, it ensures the security and privacy of data during transmission and sharing, providing more reliable and efficient technical support for photovoltaic inverter test data analysis, thereby achieving the technical effect of improving the real-time performance and privacy of data processing.

[0005] This application provides a photovoltaic inverter test data analysis system based on edge computing, comprising: a data acquisition and fusion layer, which includes a multi-source data acquisition module and an edge intelligent gateway. The multi-source data acquisition module integrates photovoltaic module sensors and an infrared imager, and acquires electrical parameters, environmental data, and thermal imaging data based on the multi-source data acquisition module. The edge intelligent gateway receives the acquired data from the multi-source data acquisition module and performs data fusion preprocessing. Each edge intelligent gateway corresponds to one photovoltaic inverter. An edge intelligent collaboration layer is used to construct an intelligent collaboration network, which includes internal collaborative communication and external federated privacy communication of the edge intelligent gateways. The intelligent collaboration network updates the prediction test channel of the edge intelligent gateways. An adaptive optimization layer is used to receive real-time response parameters of the photovoltaic inverter based on the prediction test channel and synchronously send the fusion preprocessing results to the prediction test channel for adaptation analysis of the real-time response parameters. A correction layer is used to receive the adaptation analysis results and perform control optimization of the photovoltaic inverter.

[0006] In a possible implementation, the edge intelligent collaboration layer is used to: activate internal collaborative communication of the intelligent collaboration network to obtain the equipment parameters and balanced operating environment of each photovoltaic inverter; use the equipment parameters and balanced operating environment as evaluation features to perform adaptive clustering of the photovoltaic inverters and establish multiple sets of adaptive clustering results; obtain the clustering energy values ​​of the multiple sets of adaptive clustering results, filter the multiple sets of adaptive clustering results according to the clustering energy values, and establish a unique clustering result; establish a basic prediction test channel using the unique clustering result and internal collaborative communication, and establish the prediction test channel through the basic prediction test channel.

[0007] In a possible implementation, the edge intelligent collaboration layer is used to: establish a matching database of clustering energy values; obtain the number of clusters in multiple sets of adaptive clustering results, and use the number of clusters as a first clustering energy matching coefficient; establish the cluster center of each clustering result in multiple sets of adaptive clustering results, and establish a second clustering energy matching coefficient based on the difference between the cluster center and the farthest cluster edge under the corresponding clustering result; obtain the cluster size of each clustering result in multiple sets of adaptive clustering results, and establish a third clustering matching coefficient based on the cluster size; and perform clustering energy matching of the matching database through the first clustering energy matching coefficient, the second clustering energy matching coefficient, and the third clustering matching coefficient to obtain the clustering energy values ​​of multiple sets of adaptive clustering results.

[0008] In a possible implementation, the edge intelligent collaboration layer is used to: call the cluster shape within the clustering result based on the unique clustering result, take the cluster center corresponding to the current clustering result as the starting point, and establish incremental functional requirements based on the cluster shape; use the incremental functional requirements as the matching target, establish a first database based on the internal collaborative communication, and evaluate the data in the first database to establish a data evaluation result; optimize the incremental functional requirements based on the data evaluation result, use the optimized result as the matching target, and establish a second database based on the federated privacy communication; and incrementally learn the basic prediction test channel based on the first database and the second database to establish a prediction test channel.

[0009] In a possible implementation, the system further includes: an environmental correction layer, used to predict environmental changes based on the environmental data, establish environmental change prediction results, the environmental change prediction results having a prediction reliability identifier, establish parameter retention rewards for transient control parameters based on the environmental change prediction results, and optimize the adaptation analysis results using the parameter retention rewards.

[0010] In a possible implementation, the system further includes: a joint processing layer, used to receive response parameters of the photovoltaic inverter, perform response consistency verification based on the response parameters, establish verification results, and if the verification results do not meet a preset threshold, generate a joint processing instruction, and call the joint photovoltaic inverter through the joint processing instruction to perform joint response processing.

[0011] In a possible implementation, the system further includes: a self-feedback optimization layer, used to establish control evaluation results of the photovoltaic inverter and generate control evaluation feedback, and to perform joint feedback optimization of the predictive test channel using the control evaluation feedback.

[0012] This application also provides a photovoltaic inverter test data analysis method based on edge computing, including: acquiring electrical parameters, environmental data, and thermal imaging data based on a multi-source data acquisition module, wherein the multi-source data acquisition module integrates photovoltaic module sensors and an infrared imager; receiving the acquired data from the multi-source data acquisition module based on an edge smart gateway and performing data fusion preprocessing, wherein each edge smart gateway corresponds to one photovoltaic inverter; constructing an intelligent collaborative network, wherein the intelligent collaborative network includes internal collaborative communication and external federated privacy communication of the edge smart gateway, and updating the prediction test channel of the edge smart gateway with the intelligent collaborative network; receiving the real-time response parameters of the photovoltaic inverter based on the prediction test channel, and simultaneously sending the fusion preprocessing results to the prediction test channel, and performing adaptation analysis of the real-time response parameters with the prediction test channel; receiving the adaptation analysis results and performing control optimization of the photovoltaic inverter.

[0013] The proposed edge computing-based photovoltaic inverter test data analysis system and method, as outlined in this application, first acquires electrical parameters, environmental data, and thermal imaging data using a multi-source data acquisition module integrating photovoltaic module sensors and an infrared imager. An edge smart gateway receives the acquired data from the multi-source data acquisition module and performs data fusion preprocessing. Each edge smart gateway corresponds to one photovoltaic inverter. Then, an intelligent collaborative network is constructed, including internal collaborative communication and external federated privacy communication between the edge smart gateways. This network updates the predictive test channel of the edge smart gateways. The predictive test channel then receives the real-time response parameters of the photovoltaic inverter and simultaneously sends the fusion preprocessing results to the predictive test channel for adaptation analysis of the real-time response parameters. Finally, the adaptation analysis results are received, and control optimization of the photovoltaic inverter is executed, achieving the technical effects of improving the real-time performance and privacy of data processing. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a schematic diagram of the structure of the photovoltaic inverter test data analysis system based on edge computing provided in the embodiments of this application.

[0016] Figure 2 This is a flowchart illustrating the edge computing-based photovoltaic inverter test data analysis method provided in this application embodiment.

[0017] Figure labeling: Data acquisition and fusion layer 10, edge intelligent collaboration layer 20, adaptive optimization layer 30, correction layer 40. Detailed Implementation

[0018] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0021] This application provides an edge computing-based photovoltaic inverter test data analysis system, such as... Figure 1 As shown, the system includes: The data acquisition and fusion layer 10 includes a multi-source data acquisition module and an edge intelligent gateway. The multi-source data acquisition module integrates a photovoltaic module sensor and an infrared imager. Based on the multi-source data acquisition module, electrical parameters, environmental data, and thermal imaging data are acquired. The edge intelligent gateway receives the acquired data from the multi-source data acquisition module and performs data fusion preprocessing. Each edge intelligent gateway corresponds to a photovoltaic inverter. Specifically, the data acquisition and fusion layer 10 is responsible for collecting data from multiple sources and performing preliminary processing and fusion at the edge, providing a foundation for subsequent analysis and optimization. This layer integrates the multi-source data acquisition module and the edge intelligent gateway to achieve real-time data acquisition, preprocessing, and fusion. The multi-source data acquisition module integrates various sensors to collect diverse data from the photovoltaic inverter and its surrounding environment. Specifically, photovoltaic module sensors are installed on the photovoltaic modules to collect electrical parameters such as voltage, current, and power in real time, reflecting the operating status and performance of the photovoltaic modules. Environmental sensors collect environmental data such as temperature, humidity, and light intensity. Infrared imagers use infrared radiation for non-contact temperature measurement, capturing thermal imaging data of the photovoltaic module surface to reflect the temperature distribution of the module, which is used to analyze the thermal efficiency and potential fault points of the photovoltaic modules. The edge intelligent gateway is a device deployed on edge computing nodes. It receives data from the multi-source data acquisition module via wired or wireless means and performs preliminary processing, fusion, and transmission. Specifically, it aligns data from different sensors in time to reflect the system status at the same point in time, integrates electrical parameters, environmental data, and thermal imaging data according to a certain format to form a complete data packet, and extracts useful features for analysis from the fused data, such as power fluctuation characteristics and abnormal temperature areas. Based on data fusion, further preprocessing operations such as data cleaning, compression, and encryption are performed to improve data transmission efficiency and security.

[0022] The edge intelligent collaboration layer 20 is used to construct an intelligent collaboration network. This network includes internal collaborative communication between edge intelligent gateways and external federated privacy communication, used to update the prediction test channels of the edge intelligent gateways. Specifically, the edge intelligent collaboration layer 20 is responsible for building and maintaining an intelligent collaboration network. This network achieves information sharing and collaborative work among edge intelligent gateways through internal collaborative communication (the data exchange and collaboration process between edge intelligent gateways within the same intelligent collaboration network) and external federated privacy communication mechanisms, thereby optimizing the prediction test channels and improving the overall system performance. First, the position and role of the edge intelligent gateways in the intelligent collaboration network are determined, and the protocols and specifications for internal collaboration and external communication are designed. Each edge intelligent gateway must register and authenticate before joining the intelligent collaboration network to ensure network security and reliability. Edge smart gateways share their processed pre-processed data and prediction model parameters via local area networks or dedicated communication networks (such as MQTT and CoAP). Utilizing distributed computing technology, these gateways collaboratively participate in the training process of machine learning or deep learning models, improving model accuracy and generalization ability. Prediction results from multiple gateways can be integrated through weighted averaging or voting to obtain more accurate predictions. Privacy-preserving technologies such as federated learning allow edge smart gateways in different locations to jointly train and optimize the global model without directly exchanging raw data. Each edge smart gateway periodically sends updates to its local model (such as gradients and weight differences) to the central server (or a designated coordinating node). The server updates the global model based on this information and synchronizes the updated model parameters back to each edge smart gateway. Based on the data collected in the intelligent collaborative network and the model prediction results, the performance of the prediction test channel is evaluated. Based on the evaluation results, optimization strategies are developed, such as adjusting prediction model parameters or replacing them with more suitable algorithms. These optimization strategies are applied to the prediction test channel, updating relevant configurations and model parameters in the edge smart gateways to adapt to changes in the operating status of the photovoltaic inverter.

[0023] In one possible implementation, the edge intelligent collaboration layer 20 is used to: activate the internal collaborative communication of the intelligent collaboration network to obtain the equipment parameters and balanced operating environment of each photovoltaic inverter; use the equipment parameters and balanced operating environment as evaluation features to perform adaptive clustering of the photovoltaic inverters and establish multiple sets of adaptive clustering results; obtain the clustering energy values ​​of the multiple sets of adaptive clustering results, filter the multiple sets of adaptive clustering results according to the clustering energy values, and establish a unique clustering result; establish a basic prediction test channel with the unique clustering result and the internal collaborative communication, and establish the prediction test channel through the basic prediction test channel.

[0024] Specifically, after the edge smart gateway is activated, it automatically joins the defined intelligent collaboration network. Communication links are established between edge smart gateways via network protocols (such as MQTT and CoAP), verifying the identity and permissions of each gateway to ensure network security. Internal collaborative communication services are then initiated, allowing the exchange of data and information between the gateways. The edge smart gateway collects equipment parameters (data describing the physical characteristics and performance indicators of the photovoltaic inverters, such as model, rated power, and voltage range) from each photovoltaic inverter through a multi-source data acquisition module. Simultaneously, it collects environmental data from the photovoltaic inverters (such as temperature, humidity, and light intensity) and performs equalization processing (adjusting the environmental data to a uniform scale or range for fair comparison and analysis, such as standardization and normalization) to eliminate the impact of different environments on data analysis. Clustering algorithms (such as K-means and DBSCAN) are used to perform cluster analysis on the data of all photovoltaic inverters, based on equipment parameters and balanced operating environment as evaluation features. Multiple initial clustering results are generated. Adaptive clustering optimization is performed by adjusting clustering parameters (such as the number of cluster centers and distance metrics) based on the quality of the clustering results (such as silhouette coefficient and Calinski-Harabasz index). The clustering energy value (a quantitative indicator of clustering quality, such as sum of intra-cluster variance and inter-cluster distance) of each adaptive clustering result is calculated. The clustering energy values ​​are then sorted, and the group with the optimal clustering energy value (or meeting specific conditions) is selected as the unique clustering result. This unique clustering result is verified and confirmed to ensure its representativeness and stability. Based on the unique clustering result and combined with an internal collaborative communication mechanism, a basic prediction test channel is constructed. This basic prediction test channel is a preliminary channel established for testing and verifying the prediction model. On the basic predictive test channel, the parameters and model of the predictive test channel are dynamically adjusted to improve prediction accuracy and response speed. After optimization and adjustment, the predictive test channel is obtained, which is used to actually predict the operating status of photovoltaic inverters. This implementation method, through the internal collaborative communication of the intelligent collaborative network, makes full use of the equipment parameters and operating environment data of each photovoltaic inverter to perform adaptive clustering analysis, accurately reflecting the actual operating status of the photovoltaic inverters. By screening and optimizing the clustering results, a more stable and reliable predictive test channel is established, thereby achieving the technical effect of improving the accuracy and efficiency of photovoltaic inverter test data analysis.

[0025] In one possible implementation, the edge intelligent collaboration layer 20 is used to: establish a matching database of clustering energy values; obtain the number of clusters in multiple sets of adaptive clustering results, and use the number of clusters as a first clustering energy matching coefficient; establish the cluster center of each clustering result in multiple sets of adaptive clustering results, and establish a second clustering energy matching coefficient based on the difference between the cluster center and the farthest cluster edge under the corresponding clustering result; obtain the cluster size of each clustering result in multiple sets of adaptive clustering results, and establish a third clustering matching coefficient based on the cluster size; and perform clustering energy matching of the matching database through the first clustering energy matching coefficient, the second clustering energy matching coefficient, and the third clustering matching coefficient to obtain the clustering energy values ​​of multiple sets of adaptive clustering results.

[0026] Specifically, the matching database is a database used to store and query cluster energy values ​​and their corresponding features. It contains multiple entries, each associated with a set of clustering features (such as the number of clusters, the density index between cluster centers and edges, cluster size, etc.) and a corresponding cluster energy value. The cluster energy value is obtained based on prior knowledge, experimental data, or expert evaluation. The number of clusters is extracted from each set of clustering results. For each set of clustering results, the centroid of each cluster (i.e., the mean or median of all points in that cluster) is calculated. The point furthest from the cluster center in each cluster is found, and its distance from the cluster center is calculated. A second cluster energy matching coefficient is established based on the difference between the cluster center and the furthest cluster edge (or its reciprocal, to represent density). For each cluster in each set of clustering results, the number of data points it contains, i.e., the cluster size, is calculated. A third cluster matching coefficient is established based on the cluster size (which can be used directly or transformed, such as taking the logarithm). The first, second, and third cluster energy matching coefficients are matched with entries in the matching database to find the most similar entries or those meeting specific conditions. The corresponding cluster energy values ​​are then read from the matched entries. This implementation accelerates the evaluation and selection process of clustering results by pre-establishing and storing a matching database of cluster energy values, improving overall processing efficiency. Furthermore, the cluster energy values ​​are derived from a comprehensive evaluation based on multiple dimensions of clustering features, fully reflecting the quality of the clustering results, thus achieving the technical effect of improving the accuracy of obtaining cluster energy values.

[0027] In one possible implementation, the edge intelligent collaboration layer 20 is used to: call the cluster shape within the clustering result based on the unique clustering result, take the cluster center corresponding to the current clustering result as the starting point, and establish incremental functional requirements based on the cluster shape; use the incremental functional requirements as the matching target, establish a first database based on the internal collaborative communication, and evaluate the data in the first database to establish a data evaluation result; optimize the incremental functional requirements based on the data evaluation result, use the optimized result as the matching target, and establish a second database based on the federated privacy communication; and incrementally learn the basic prediction test channel based on the first database and the second database to establish a prediction test channel.

[0028] Specifically, for each cluster in the unique clustering result, its shape characteristics, such as circular, elliptical, or linear, are identified by calculating statistical measures such as the distribution, density, and distance of points within the cluster. The center of each cluster is used as the starting point for analysis or modeling. Based on the cluster shape, the characteristics of the entities or phenomena that the cluster may represent are analyzed, such as distribution range and changing trends. Based on the analysis results, incremental functional requirements for that cluster are proposed. These are new functional or optimization requirements aimed at improving system performance, based on the current clustering results, including further data collection, processing, analysis, or prediction. Through the communication mechanism within the edge intelligent collaboration layer, the incremental functional requirements are sent to the relevant data processing or storage units. Based on these requirements, relevant data is collected, organized, and stored to form a first database. The data in the first database undergoes quality assessment and integrity checks to ensure its reliability and validity. The conclusions and problems discovered during the evaluation process are compiled into data evaluation results. Based on data evaluation results, incremental functional requirements are adjusted and optimized to ensure they match the actual data situation. Utilizing privacy-preserving technologies such as federated learning, communication and data exchange with other edge intelligent nodes or data centers are conducted while protecting data privacy. Based on the optimized incremental functional requirements, a wider range of data is collected through federated privacy communication to establish a second database. Using data from both the first and second databases, incremental learning is performed on the basic prediction test channel—that is, the prediction model is continuously updated and optimized. After incremental learning, a more accurate and reliable prediction model is obtained and applied to real-world scenarios to establish the prediction test channel. This implementation method establishes new prediction test channels by incrementally learning from the basic prediction test channel, ensuring that the model can continuously adapt to new data features, patterns, or trends, achieving the technical effect of improving the prediction accuracy of the prediction test channel and supporting rapid iteration.

[0029] The adaptive optimization layer 30 receives real-time response parameters from the photovoltaic inverter via the predictive test channel and simultaneously sends the fusion preprocessing results to the predictive test channel for adaptation analysis of the real-time response parameters. Specifically, the adaptive optimization layer 30 first receives data from the data acquisition and fusion layer 10 after fusion preprocessing and simultaneously acquires the latest status and parameters of the predictive test channel (located in the edge intelligent collaboration layer 20). Using the predictive model (which can be a machine learning model or a deep learning model, etc.) within the predictive test channel, it performs adaptation analysis on the real-time response parameters (various performance index data generated by the photovoltaic inverter during real-time operation, such as voltage, current, power output, etc.), including comparing the differences between the actual response and the predicted response, identifying outliers or trends, and evaluating the performance status of the photovoltaic inverter. Based on the results of the adaptation analysis, the adaptive optimization layer 30 generates corresponding optimization strategies. These optimization strategies are measures or schemes used to improve the performance of the photovoltaic inverter or reduce the risk of failure, including adjusting the operating parameters of the photovoltaic inverter (such as voltage range, current limit), optimizing the control algorithm, or triggering preventive maintenance measures.

[0030] The calibration layer 40 receives adaptation analysis results and performs control optimization of the photovoltaic inverter. Specifically, the calibration layer 40 receives adaptation analysis results from the adaptive optimization layer 30 via API interfaces or message queues. These results include evaluations of the photovoltaic inverter's real-time response parameters, identifying potential performance issues or optimization opportunities. A data parsing engine analyzes the received adaptation analysis results, understanding the optimization suggestions or control commands, and generates specific optimization strategies based on these suggestions using a strategy generation algorithm. The optimization strategies are then translated into specific control commands and sent to the corresponding photovoltaic inverter via appropriate communication protocols, such as Modbus or OPC UA, for remote control of the photovoltaic inverter's operating parameters or working mode. After sending the control commands, the calibration layer 40 monitors the photovoltaic inverter's execution status to ensure correct execution and collects post-execution feedback data. Based on the post-execution feedback data, the calibration layer evaluates the effectiveness of the optimization strategy, determines whether the optimization has achieved the expected goals, and feeds the evaluation results back to the adaptive optimization layer 30 for subsequent iterative optimization. This application employs techniques such as directly collecting electrical parameters, environmental data, and thermal imaging data near the photovoltaic inverter, performing data fusion and preprocessing through an edge smart gateway to reduce data transmission volume, constructing an intelligent collaborative network, and enabling internal collaborative communication between edge smart gateways and external federated privacy communication. This ensures data sharing and collaboration while protecting privacy. The application also dynamically updates the predictive test channel based on real-time data, receives real-time response parameters from the photovoltaic inverter based on the predictive test channel, and simultaneously sends fusion preprocessing results. Real-time response parameter adaptation analysis is performed, and based on the adaptation analysis results, control optimization of the photovoltaic inverter is executed to ensure the photovoltaic inverter always operates in its optimal state. By pushing data processing capabilities down to the vicinity of the data source, i.e., the edge node where the photovoltaic inverter is located, real-time data acquisition, preprocessing, and analysis are achieved, significantly improving the real-time performance of data processing. Combined with encryption technology and privacy protection mechanisms, the security and privacy of data during transmission and sharing are ensured, providing more reliable and efficient technical support for photovoltaic inverter test data analysis, thereby achieving the technical effect of improving the real-time performance and privacy of data processing.

[0031] In one possible implementation, the system further includes: an environmental correction layer, used to predict environmental changes based on the environmental data, establish environmental change prediction results, the environmental change prediction results having a prediction reliability identifier, establish parameter retention rewards for transient control parameters based on the environmental change prediction results, and optimize the adaptation analysis results using the parameter retention rewards.

[0032] Specifically, the environmental correction layer receives environmental data from the data acquisition and fusion layer 10 via an edge intelligent gateway. This data includes temperature, humidity, wind speed, and light intensity. Time series analysis, machine learning models, or prediction algorithms are used to process and analyze the collected environmental data to predict environmental change trends over a future period. The confidence interval of the predicted values ​​is calculated as a quantitative indicator of prediction reliability, and a reliability label is assigned to the prediction results to represent their reliability or confidence level. Based on the environmental change prediction results, a parameter retention reward mechanism is established for the transient control parameters of the photovoltaic inverter (such as voltage regulation speed and current limiting). This mechanism aims to optimize the performance of the photovoltaic inverter by maintaining or adjusting the control parameters when the environment changes. For example, reinforcement learning algorithms are used to learn how to adjust the control parameters to maximize performance rewards based on the environmental change prediction results and the real-time response of the photovoltaic inverter. The parameter retention rewards of the transient control parameters generated by the environmental correction layer are incorporated into the adaptation analysis process of the adaptive optimization layer 30 to optimize the final adaptation analysis results. This implementation method, through a parameter retention reward mechanism, enables the system to maintain the stable operation of the photovoltaic inverter while ensuring performance optimization, reducing system fluctuations or failures that may be caused by frequent adjustments to control parameters, thus achieving the technical effect of enhancing system stability.

[0033] In one possible implementation, the system further includes: a joint processing layer, used to receive response parameters of the photovoltaic inverter, perform response consistency verification based on the response parameters, establish a verification result, and if the verification result does not meet a preset threshold, generate a joint processing instruction, and call the joint photovoltaic inverter through the joint processing instruction to perform joint response processing.

[0034] Specifically, the joint processing layer receives real-time response parameters from the photovoltaic inverter, including output power, voltage, current, and temperature. It performs consistency verification on the received response parameters, checking whether they conform to the expected response pattern or range. For example, it sets a series of predefined rules based on the photovoltaic inverter design specifications and operation manual, such as power fluctuation range and voltage / current stability requirements. It then performs statistical analysis on the real-time response parameters, such as calculating the average, standard deviation, maximum, and minimum values, and compares these values ​​with the predefined rules. Anomaly detection algorithms (such as distance-based anomaly detection and density-based anomaly detection) are used to identify abnormal response parameters. Based on the response consistency verification results, a verification report or verification result identifier is generated, including a list of parameters that passed verification, details of parameters that failed verification, and the corresponding anomaly types. The layer evaluates whether the verification results meet preset thresholds or conditions. These preset thresholds can be empirical values ​​derived from historical data statistical analysis or fixed values ​​set based on the photovoltaic inverter performance requirements. The verification results are compared with the preset thresholds to determine whether the photovoltaic inverter response is normal. If the verification result fails to meet the preset threshold, indicating an abnormal or inconsistent response from the photovoltaic inverter, the joint processing layer generates joint processing instructions. These instructions include specific operational commands such as adjusting photovoltaic inverter control parameters, activating backup photovoltaic inverters, and switching operating modes. Through these joint processing instructions, one or more photovoltaic inverters are invoked to participate in the response processing, utilizing the communication protocols and interfaces between the photovoltaic inverters. The joint photovoltaic inverters then execute corresponding response processing operations based on the joint processing instructions to restore or optimize the photovoltaic inverter's operating state. These operations include load balancing adjustments, fault isolation and recovery, and power output optimization. This implementation, by invoking multiple photovoltaic inverters through the joint processing layer, achieves advanced functions such as load balancing and fault isolation, thereby enhancing the system's ability to handle complex situations.

[0035] In one possible implementation, the system further includes: a self-feedback optimization layer, used to establish control evaluation results of the photovoltaic inverter and generate control evaluation feedback, and to perform joint feedback optimization of the predictive test channel using the control evaluation feedback.

[0036] Specifically, the self-feedback optimization layer is responsible for generating feedback based on the control evaluation results and continuously optimizing the performance of the photovoltaic inverter through a joint feedback optimization mechanism. The self-feedback optimization layer receives real-time response parameters from the adaptive optimization layer 30 and control optimization results from the correction layer 40. It also receives prediction results and performance evaluations generated from real-time data in the predictive test channel. Based on this data, the self-feedback optimization layer constructs control evaluation results for the photovoltaic inverter, including assessing the effectiveness of the current control strategy, identifying potential problem areas, and predicting future performance. According to the control evaluation results, the self-feedback optimization layer generates specific control evaluation feedback, including suggestions for adjusting control parameters, directions for optimizing control algorithms, or instructions for hardware maintenance. The generated control evaluation feedback is sent to the predictive test channel to guide the adjustment and optimization of the prediction model or algorithm in that channel. The predictive test channel adjusts its internal parameters or updates its model based on the feedback to improve the accuracy of predicting the future operating state of the photovoltaic inverter and the effectiveness of the optimization suggestions. Through this joint feedback optimization approach, the predictive test channel can continuously learn and improve to better support the real-time optimization and long-term performance improvement of the photovoltaic inverter. The feedback optimization process between the self-feedback optimization layer and the predictive test channel is an iterative process. With the continuous input of new data and the ongoing optimization of the predictive model, the performance of the entire system will continuously improve. During this process, the adaptive optimization layer 30 and the correction layer 40 will also make corresponding adjustments and optimizations based on the latest control evaluation feedback to ensure that the photovoltaic inverter always operates in its optimal state. This implementation method, by introducing a self-feedback optimization layer, constructs a closed-loop optimization and feedback mechanism. Through this mechanism, the system can evaluate the operating status and control effect of the photovoltaic inverter in real time and automatically generate optimization suggestions based on the evaluation results. These suggestions are verified and adjusted through the predictive testing channel, ultimately transforming into actual optimizations of the photovoltaic inverter control strategy. This approach not only improves the operating efficiency and stability of the photovoltaic inverter but also enhances the system's adaptability and intelligence. Simultaneously, the joint work of the predictive testing channel and the self-feedback optimization layer promotes the full utilization of data and the continuous accumulation of knowledge, providing strong support for the long-term development and performance improvement of the system.

[0037] In the above text, refer to Figure 1 A photovoltaic inverter test data analysis system based on edge computing according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 This invention describes a photovoltaic inverter test data analysis method based on edge computing according to an embodiment of the present invention.

[0038] The edge computing-based photovoltaic inverter test data analysis method according to embodiments of the present invention addresses the technical problems of insufficient real-time performance and privacy protection in existing photovoltaic inverter test data analysis methods, thereby improving the real-time performance and privacy of data processing. The edge computing-based photovoltaic inverter test data analysis method includes: Electrical parameters, environmental data, and thermal imaging data are acquired using a multi-source data acquisition module, which integrates photovoltaic module sensors and an infrared imager. The data is received from the multi-source data acquisition module via an edge smart gateway, which performs data fusion preprocessing. Each edge smart gateway corresponds to one photovoltaic inverter. An intelligent collaborative network is constructed, including internal collaborative communication and external federated privacy communication between the edge smart gateways. This network is used to update the prediction test channel of the edge smart gateways. Real-time response parameters of the photovoltaic inverters are received via the prediction test channel, and the fusion preprocessing results are simultaneously sent to the prediction test channel for adaptation analysis of the real-time response parameters. The adaptation analysis results are received, and control optimization of the photovoltaic inverters is executed.

[0039] The step of updating the prediction test channel of the edge intelligent gateway using the intelligent collaborative network further includes: activating the internal collaborative communication of the intelligent collaborative network to obtain the device parameters and balanced operating environment of each photovoltaic inverter; using the device parameters and balanced operating environment as evaluation features to perform adaptive clustering of the photovoltaic inverters and establish multiple sets of adaptive clustering results; obtaining the clustering energy values ​​of the multiple sets of adaptive clustering results, filtering the multiple sets of adaptive clustering results based on the clustering energy values, and establishing a unique clustering result; establishing a basic prediction test channel using the unique clustering result and the internal collaborative communication, and establishing the prediction test channel through the basic prediction test channel.

[0040] The step of obtaining clustering energy values ​​of multiple sets of adaptive clustering results further includes: establishing a matching database of clustering energy values; obtaining the number of clusters in multiple sets of adaptive clustering results, and using the number of clusters as a first clustering energy matching coefficient; establishing the cluster center of each clustering result in multiple sets of adaptive clustering results, and establishing a second clustering energy matching coefficient based on the difference between the cluster center and the farthest cluster edge under the corresponding clustering result; obtaining the cluster size of each clustering result in multiple sets of adaptive clustering results, and establishing a third clustering matching coefficient based on the cluster size; and performing clustering energy matching in the matching database using the first clustering energy matching coefficient, the second clustering energy matching coefficient, and the third clustering matching coefficient to obtain the clustering energy values ​​of multiple sets of adaptive clustering results.

[0041] The step of establishing the prediction test channel through the basic prediction test channel further includes: calling the cluster shape within the clustering result based on the unique clustering result, taking the cluster center corresponding to the current clustering result as the starting point, and establishing incremental functional requirements based on the cluster shape; using the incremental functional requirements as the matching target, establishing a first database based on the internal collaborative communication, and evaluating the data in the first database to establish a data evaluation result; optimizing the incremental functional requirements based on the data evaluation result, using the optimized result as the matching target, and establishing a second database based on the federated privacy communication; and incrementally learning the basic prediction test channel based on the first database and the second database to establish the prediction test channel.

[0042] The method further includes: predicting environmental changes based on the environmental data, establishing environmental change prediction results, wherein the environmental change prediction results are accompanied by a prediction reliability identifier, establishing a parameter retention reward for transient control parameters based on the environmental change prediction results, and optimizing the adaptation analysis results using the parameter retention reward.

[0043] The method further includes: receiving response parameters from a photovoltaic inverter, performing response consistency verification based on the response parameters, establishing a verification result, and generating a joint processing instruction if the verification result does not meet a preset threshold, and calling the joint photovoltaic inverter through the joint processing instruction to perform joint response processing.

[0044] The method further includes: establishing control evaluation results for the photovoltaic inverter and generating control evaluation feedback, and using the control evaluation feedback to perform joint feedback optimization of the predictive test channel.

[0045] The edge computing-based photovoltaic inverter test data analysis system provided in this embodiment of the invention can execute the edge computing-based photovoltaic inverter test data analysis method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0046] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0047] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A photovoltaic inverter test data analysis system based on edge computing, characterized in that, The system comprises: a data acquisition and fusion layer, comprising a multi-source data acquisition module and an edge intelligent gateway, the multi-source data acquisition module is integrated with a photovoltaic component sensor and an infrared imager, electrical parameters, environmental data and thermal imaging data are obtained based on the multi-source data acquisition module, the edge intelligent gateway receives the collected data of the multi-source data acquisition module, and the fusion preprocessing of the data is performed, wherein each edge intelligent gateway corresponds to one photovoltaic inverter; an edge intelligent collaboration layer for building an intelligent collaboration network, the intelligent collaboration network comprises internal collaborative communication and external federal privacy communication of the edge intelligent gateway, and the intelligent collaboration network is used to update the prediction test channel of the edge intelligent gateway; an adaptive optimization layer for receiving real-time response parameters of the photovoltaic inverter based on the prediction test channel, and synchronously sending the fusion preprocessing result to the prediction test channel, so that the prediction test channel performs adaptive analysis on the real-time response parameters; a correction layer for receiving the adaptive analysis result and performing control optimization of the photovoltaic inverter.

2. The edge-computing-based photovoltaic inverter test data analysis system of claim 1, wherein, The edge intelligent collaboration layer is used to: activate the internal collaborative communication of the intelligent collaboration network, obtain device parameters and balanced working environment of each photovoltaic inverter; use the device parameters and balanced working environment as evaluation features to perform adaptive clustering of the photovoltaic inverters, and establish a plurality of adaptive clustering results; obtain clustering energy values of the plurality of adaptive clustering results, perform adaptive clustering result screening according to the clustering energy values, and establish a unique clustering result; establish a basic prediction test channel based on the unique clustering result and the internal collaborative communication, and establish the prediction test channel through the basic prediction test channel.

3. The edge-computing-based photovoltaic inverter test data analysis system of claim 2, wherein, The edge intelligent collaboration layer is used to: establish a matching database of the clustering energy values; obtain a clustering number in the plurality of adaptive clustering results, and use the clustering number as a first clustering energy matching coefficient; establish a clustering center of each adaptive clustering result in the plurality of adaptive clustering results, and establish a second clustering energy matching coefficient according to a difference between the clustering center and a farthest clustering edge under the corresponding adaptive clustering result; obtain a clustering size of each adaptive clustering result in the plurality of adaptive clustering results, and establish a third clustering matching coefficient based on the clustering size; perform clustering energy matching of the matching database through the first clustering energy matching coefficient, the second clustering energy matching coefficient and the third clustering matching coefficient, and obtain the clustering energy values of the plurality of adaptive clustering results.

4. The edge-computing-based photovoltaic inverter test data analysis system of claim 2, wherein, The edge intelligent collaboration layer is used to: call a clustering shape in the unique clustering result according to the unique clustering result, use a clustering center corresponding to a current clustering result as a starting point, and establish an incremental functional requirement based on the clustering shape; use the incremental functional requirement as a matching target, establish a first database based on the internal collaborative communication, perform data evaluation on the first database, and establish a data evaluation result; after optimizing the incremental functional requirement based on the data evaluation result, use an optimized result as a matching target, establish a second database based on the federal privacy communication; perform incremental learning on the basic prediction test channel according to the first database and the second database, and establish the prediction test channel.

5. The edge-computing-based photovoltaic inverter test data analysis system of claim 1, wherein, The system further comprises: An environment correction layer is configured to perform environment change prediction based on the environment data, establish an environment change prediction result, and label the environment change prediction result with a prediction credibility identifier. A parameter retention reward of the transient control parameter is established based on the environment change prediction result, and the parameter retention reward is used to optimize the adaptation analysis result.

6. The edge-computing-based photovoltaic inverter test data analysis system of claim 1, wherein, The system further comprises: A joint processing layer is configured to receive a response parameter of the photovoltaic inverter, perform response consistency verification based on the response parameter, establish a verification result, and generate a joint processing instruction if the verification result does not meet a preset threshold. The joint photovoltaic inverter is called through the joint processing instruction, and joint response processing is performed.

7. The edge-computing-based photovoltaic inverter test data analysis system of claim 1, wherein, The system further comprises: A self-feedback optimization layer is configured to establish a control evaluation result of the photovoltaic inverter and generate a control evaluation feedback. The joint feedback optimization of the prediction test channel is performed based on the control evaluation feedback.

8. A method for photovoltaic inverter test data analysis based on edge computing, characterized in that, The method is implemented by the photovoltaic inverter test data analysis system based on edge computing according to any one of claims 1-7, and the method comprises: The electrical parameters, the environment data, and the thermal imaging data are obtained based on the multi-source data acquisition module. The multi-source data acquisition module is integrated with a photovoltaic module sensor and an infrared imager. The edge intelligent gateway receives the collected data of the multi-source data acquisition module and performs data fusion preprocessing. Each edge intelligent gateway corresponds to one photovoltaic inverter. An intelligent collaboration network is constructed, which includes internal collaboration communication and external federal privacy communication of the edge intelligent gateway. The prediction test channel of the edge intelligent gateway is updated based on the intelligent collaboration network. Real-time response parameters of the photovoltaic inverter are received based on the prediction test channel, and the fusion preprocessing result is synchronously sent to the prediction test channel. The real-time response parameters are adaptively analyzed based on the prediction test channel. The adaptation analysis result is received, and the control optimization of the photovoltaic inverter is performed.

Images