An Adaptive Fault Diagnosis Method and Device for a Photovoltaic Inverter

Abstract

The present invention discloses an adaptive fault diagnosis method and device for a photovoltaic inverter, which relates to the technical field of photovoltaic new energy. The method includes: calling the device operation and maintenance data of the photovoltaic inverter, extracting K fault operation data sets, and constructing an inverter fault identification model; collecting multiple fault time nodes of multiple pseudo-faults, traversing the system operation and maintenance data, obtaining multiple system fault device sets, and constructing a fault troubleshooting model; collaborating the inverter fault identification model and the fault troubleshooting model to perform real-time operation data analysis of the photovoltaic inverter for adaptive fault diagnosis. The present invention solves the technical problem in the prior art that the collaborative operation equipment may cause the photovoltaic inverter to be in a pseudo-fault state, resulting in false alarms, false shutdowns, etc., and achieves the technical effect of effectively distinguishing the true faults and pseudo-faults of the photovoltaic inverter, reducing false alarms and false shutdowns, and improving the overall reliability of the photovoltaic system.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic new energy, and particularly relates to an adaptive fault diagnosis method and device for a photovoltaic inverter. Background Art

[0002] A photovoltaic inverter is a key device in a photovoltaic power generation system, responsible for converting the direct current generated by photovoltaic modules into alternating current and transmitting it to the power grid or supplying it to a load for use. To ensure the efficient operation and safety of the photovoltaic system, fault diagnosis of the inverter is of great significance.

[0003] However, the photovoltaic inverter operates in coordination with other devices in the photovoltaic power grid. The abnormal operation of other devices may cause the photovoltaic inverter to be in a pseudo-fault state, resulting in a decrease in the accuracy of fault diagnosis, causing false alarms, false shutdowns, etc., leading to unnecessary maintenance and increasing the operation and maintenance costs of the photovoltaic power generation system. Summary of the Invention

[0004] This application provides an adaptive fault diagnosis method and device for a photovoltaic inverter, which is used to solve the technical problem that the coordinated operation devices in the prior art may cause the photovoltaic inverter to be in a pseudo-fault state, resulting in false alarms, false shutdowns, etc.

[0005] In the first aspect of this application, an adaptive fault diagnosis method for a photovoltaic inverter is provided. The method includes: calling the device operation and maintenance data of the photovoltaic inverter, and extracting K fault operation data sets of K fault types from the device operation and maintenance data based on the fault types; constructing and generating an inverter fault recognition model based on the K fault types and the K fault operation data sets; collecting multiple fault time nodes of multiple pseudo-faults from the device operation and maintenance data; calling the system operation and maintenance data of the photovoltaic power generation system, and traversing the system operation and maintenance data with the multiple fault time nodes to obtain multiple system fault device sets; constructing and generating a fault troubleshooting model based on the K fault types and the multiple system fault device sets; analyzing the real-time operation data obtained by monitoring the photovoltaic inverter through the coordinated operation of the inverter fault recognition model and the fault troubleshooting model to perform adaptive fault diagnosis on the photovoltaic inverter.

[0006] In a second aspect of the present application, an adaptive fault diagnosis device for a photovoltaic inverter is provided. The device includes: a fault operation data extraction module configured to call the device operation and maintenance data of the photovoltaic inverter and extract K fault operation data sets of K fault types from the device operation and maintenance data based on the fault types; an inverter fault identification model construction module configured to construct an inverter fault identification model based on the K fault types and the K fault operation data sets; a fault time node acquisition module configured to collect multiple fault time nodes of multiple pseudo-faults from the device operation and maintenance data; a system fault device collection module configured to call the system operation and maintenance data of the photovoltaic power generation system and traverse the system operation and maintenance data using the multiple fault time nodes to obtain multiple system fault device sets; a fault troubleshooting model construction module configured to construct a fault troubleshooting model based on the K fault types and the multiple system fault device sets; and an adaptive fault diagnosis module configured to perform data analysis on the real-time operation data obtained by monitoring the photovoltaic inverter by collaborating the inverter fault identification model and the fault troubleshooting model to perform adaptive fault diagnosis on the photovoltaic inverter.

[0007] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] An adaptive fault diagnosis method and device for a photovoltaic inverter provided in the present application relate to the technical field of photovoltaic new energy. By calling the device operation and maintenance data of the photovoltaic inverter, constructing an inverter fault identification model, collecting data by traversing the system operation and maintenance data based on multiple fault time nodes of multiple pseudo-faults, and constructing a fault troubleshooting model, and finally collaborating the inverter fault identification model and the fault troubleshooting model to perform data analysis on the real-time operation of the photovoltaic inverter for adaptive fault diagnosis, it solves the technical problem that collaborative operation devices in the prior art may cause the photovoltaic inverter to be in a pseudo-fault state, resulting in false alarms, false shutdowns, etc., and achieves the technical effect of effectively distinguishing true faults and pseudo-faults of the photovoltaic inverter, reducing false alarms and false shutdowns, and improving the overall reliability of the photovoltaic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0010] Figure 1Schematic flow chart of an adaptive fault diagnosis method for a photovoltaic inverter provided by an embodiment of the present application;

[0011] Figure 2 Schematic flow chart of generating a fault troubleshooting model in an adaptive fault diagnosis method for a photovoltaic inverter provided by an embodiment of the present application;

[0012] Figure 3 Schematic structural diagram of an adaptive fault diagnosis device for a photovoltaic inverter provided by an embodiment of the present application.

[0013] Explanation of reference numerals: Fault operation data extraction module 11, Inverter fault identification model construction module 12, Fault time node acquisition module 13, System fault equipment acquisition module 14, Fault troubleshooting model construction module 15, Adaptive fault diagnosis module 16. Detailed implementation manners

[0014] The present application provides an adaptive fault diagnosis method and device for a photovoltaic inverter, which are used to solve the technical problems in the prior art that co - operating devices may cause the photovoltaic inverter to be in a pseudo - fault state, resulting in false alarms, false shutdowns, etc.

[0015] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0016] It should be noted that the terms "first", "second", etc. in the specification and the above - mentioned accompanying drawings of the present application are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non - exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices.

[0017] Embodiment 1, as Figure 1 shown, the present application provides an adaptive fault diagnosis method for a photovoltaic inverter, and the method includes:

[0018] P10: Call the device operation and maintenance data of the photovoltaic inverter, and extract K fault operation data sets of K fault types from the device operation and maintenance data based on the fault type.

[0019] Specifically, first call the device operation and maintenance data of the photovoltaic inverter. These operation and maintenance data usually include various key parameters collected during the actual operation of the inverter, such as input voltage, output voltage, current, power, temperature, etc. These data are real records of the operation performance of the photovoltaic inverter under different working conditions and can reflect the health status of the device.

[0020] Next, classify and organize these operation and maintenance data based on the fault type. The fault type refers to various abnormal operation situations that the inverter may encounter, such as overvoltage, undervoltage, overcurrent, overtemperature, inverter module failure, etc. Assume there are K known fault types, which may be summarized based on past experience or identified as common or typical faults through an expert system or machine learning algorithm. Each fault type corresponds to specific fault characteristics.

[0021] For each fault type, the system will screen out the relevant operation data from the huge operation and maintenance data to form the corresponding fault operation data set. To achieve this process, a series of data processing and analysis techniques are required. For example, data cleaning techniques can be used to remove noise data and incomplete records to ensure the accuracy and reliability of the data. Data classification techniques, such as clustering analysis or decision tree algorithms, can be used to automatically classify the operation and maintenance data according to the fault type. At the same time, domain knowledge can also be combined to verify and correct some data through manual annotation to improve the accuracy of classification.

[0022] Finally, after the above processing, K fault operation data sets are output. Each data set corresponds to a specific fault type and contains information such as the operation characteristics, fault manifestations, and possible fault causes of the inverter under this fault type. These data sets will serve as important inputs for constructing the inverter fault identification model in the following and provide strong data support for subsequent fault diagnosis.

[0023] P20: Build and generate an inverter fault identification model based on the K fault types and the K fault operation data sets.

[0024] Furthermore, step P20 of the embodiment of the present application further includes:

[0025] P21: Preset the sample size requirement. When any one of the K faulty operation datasets does not meet the sample size requirement, interactively obtain the photovoltaic system attributes of the photovoltaic power generation system; P22: Based on the photovoltaic system attributes, call and obtain the operation and maintenance data of M sample inverters from a pre-constructed power generation system information library, where M is a positive integer; P23: Update the K faulty operation datasets with the M sample operation and maintenance data to obtain K updated operation datasets; P24: Construct a standard fault identification model based on the gradient boosting machine algorithm; P25: Use the K updated operation datasets as training data to train and tune the parameters of the standard fault identification model to obtain K fault identification branches; P26: Parallelize the K fault identification branches to complete the construction of the inverter fault identification model. Among them, the photovoltaic system attributes include the target installation capacity, target conversion efficiency, target system efficiency, target performance ratio, target sunshine hours, and target radiation level.

[0026] It should be understood that based on the K types of fault types and K faulty operation datasets extracted in the previous step, a fault identification model of the photovoltaic inverter is constructed. Specifically, first, preset the sample size requirement, that is, the amount of data required to train the model to a preset accuracy, which can be obtained by a reasonable threshold based on the experience and theoretical calculation of model training. If, when extracting the initial data, it is found that the sample size of any one of the K faulty operation datasets does not meet this requirement, the system will automatically trigger the data supplement mechanism. At this time, the system will interact with the user or an external system to obtain the photovoltaic system attributes of the photovoltaic power generation system, which include but are not limited to the target installation capacity, target conversion efficiency, target system efficiency, target performance ratio, target sunshine hours, and target radiation level, etc., which jointly describe the basic characteristics and operating environment of the photovoltaic power generation system.

[0027] Next, based on the obtained photovoltaic system attributes, call the operation and maintenance data of multiple (assumed to be M) sample inverters from the pre-constructed power generation system information library. The power generation system information library is a pre-stored database that contains the operation and maintenance data of multiple sample photovoltaic inverters. These sample inverters have a certain similarity in attributes with the current photovoltaic power generation system, so their operation and maintenance data can be used as supplementary data. By calling these sample data, the sample size of the faulty operation dataset can be increased, and the diversity and representativeness of the data can be improved.

[0028] Further, the operation and maintenance data of the M sample inverters called are used to update the original K fault operation data sets. This update process may include operations such as data merging, screening, and preprocessing to ensure that the updated data sets contain a sufficient sample size while maintaining data consistency and accuracy. Then, the updated K fault operation data sets are used as training data and input into the standard fault identification model for model training and parameter tuning. The parameters of the model are continuously optimized and adjusted to improve the identification accuracy and generalization ability of the model. Finally, K fault identification branches for different fault types are obtained.

[0029] Finally, the K fault identification branches are connected in parallel to complete the construction of the inverter fault identification model. The inverter fault identification model can simultaneously process multiple fault types, comprehensively monitor and analyze the real-time operation data of the photovoltaic inverter. Once abnormal or fault signs are detected, it can quickly locate and give corresponding diagnostic results. In this way, the adaptive identification and rapid response of photovoltaic inverter faults are realized.

[0030] Further, step P22 of the embodiment of the present application further includes:

[0031] P22-1: Interactively obtain multiple sample power generation systems with the same photovoltaic panel type as the photovoltaic power generation system, and call multiple sample system attributes and multiple sample operation and maintenance data of the multiple sample power generation systems; P22-2: Based on the knowledge graph, associatively store the multiple sample power generation systems, multiple sample system attributes, and multiple sample operation and maintenance data to obtain the power generation system information library; P22-3: Calculate multiple attribute similarity coefficients between the photovoltaic system attributes and the multiple sample system attributes based on the Euclidean distance and the custom weighted distance; P22-4: Preset an attribute similarity threshold, and call the M sample operation and maintenance data of the M sample power generation systems whose attribute similarity coefficients meet the attribute similarity threshold from the power generation system information library, where the M sample operation and maintenance data are data records of the operation and maintenance of the M sample inverters in the M sample power generation systems.

[0032] Optionally, to ensure that the selected sample data highly matches the characteristics of the actual photovoltaic power generation system, the process of calling sample data can be, first, interact with the user or an external database to obtain information on multiple sample power generation systems with the same photovoltaic panel type as the current photovoltaic power generation system. These sample power generation systems are not only the same in terms of photovoltaic panel type but may also be similar in terms of geographical location, climate conditions, installation methods, etc., thus providing valuable references for subsequent data calling and model training. Then, call multiple sample system attributes (such as installed capacity, conversion efficiency, system efficiency, etc.) and multiple sample operation and maintenance data (i.e., operation records of inverters and other related equipment) of these sample power generation systems.

[0033] Furthermore, in order to effectively manage and utilize the multiple sample system attributes and multiple sample operation and maintenance data, knowledge graph technology is adopted to associatively store the multiple sample power generation systems, multiple sample system attributes, and multiple sample operation and maintenance data. A knowledge graph is a structured data representation method that can clearly show the relationships between entities and support efficient querying and reasoning. By converting the sample data into the form of a knowledge graph, it is easier to organize, retrieve, and analyze the data, thereby constructing a comprehensive power generation system information database.

[0034] Furthermore, after obtaining the photovoltaic system attributes of the current photovoltaic power generation system, the system needs to find similar sample power generation systems. For this purpose, the Euclidean distance and the custom weighted distance are used as similarity measurement criteria to calculate multiple attribute similarity coefficients between the current photovoltaic system attributes and multiple sample system attributes. The Euclidean distance is a commonly used distance measurement method that can reflect the straight-line distance between two points in a multi-dimensional space; while the custom weighted distance allows the system to perform weighted processing according to the importance of different attributes, making the similarity calculation more in line with the actual situation. By combining these two distance calculation methods, the system can more accurately evaluate the similarity between different sample power generation systems and the current photovoltaic power generation system.

[0035] Finally, after presetting an attribute similarity threshold, M sample operation and maintenance data of M sample power generation systems whose attribute similarity coefficients meet the attribute similarity threshold are called from the power generation system information database. These sample operation and maintenance data are data records of the operation and maintenance of M sample inverters in M sample power generation systems, containing rich operation status and fault information. By combining these data as a supplementary data set with the original fault operation data set, for example, based on the fault type, K sample operation data sets of K fault types are extracted from the sample operation and maintenance data, and the K sample operation data sets are mapped to update the data of the K fault operation data sets to obtain the K updated operation data sets, which can further improve the training effect and prediction accuracy of the inverter fault identification model.

[0036] P30: Multiple fault time nodes of multiple false faults are obtained by collecting from the device operation and maintenance data.

[0037] Specifically, the fault time nodes of multiple false faults are identified and extracted from the device operation and maintenance data. The false fault, also known as false alarm or false warning, refers to the situation where the system wrongly reports that the device has a fault during actual operation, while in fact the device may be in normal operation or only has minor abnormalities that do not affect the overall function, as well as false faults caused by external factors or other device abnormalities. The identification of such fault time nodes is crucial for improving the accuracy and reliability of the fault diagnosis system.

[0038] Exemplarily, first, comprehensively scan and analyze various operating parameters from the device operation and maintenance data of the photovoltaic inverter. These parameters include input voltage, output voltage, current, power, temperature, frequency, etc., which are the true reflections of the photovoltaic inverter under different operating conditions. During the normal operation of the device, any abnormal fluctuations or deviations may be recorded as fault events.

[0039] In actual operation, through data mining and anomaly detection techniques, these false faults can be identified from a large amount of device operation and maintenance data. Specifically, by using anomaly detection algorithms (such as statistical control chart methods, anomaly detection models in machine learning, etc.), the time series of the data can be effectively analyzed, and those time points that are inconsistent with the normal operation mode can be identified. To ensure the accuracy and efficiency of the identification, domain knowledge and expert experience can also be combined to optimize the algorithm parameters and identification logic. For example, according to the operating characteristics and common fault modes of the photovoltaic inverter, adjust the thresholds and sensitivities of anomaly detection to reduce false alarms and missed detections.

[0040] Once these abnormal points are identified, the system will further verify whether these time points correspond to known external interferences or abnormal states of other devices. After this multiple verification, these time points are finally determined as the fault time nodes of false faults. These nodes record the exact time when the false faults occur and provide a clear reference basis for subsequent analysis.

[0041] P40: Invoke the system operation and maintenance data of the photovoltaic power generation system, and traverse the system operation and maintenance data using the multiple fault time nodes to obtain multiple system fault device sets.

[0042] In a possible embodiment of the present application, by invoking the system operation and maintenance data of the photovoltaic power generation system and combining the multiple false fault time nodes identified in the previous steps, traverse and analyze the system operation and maintenance data, so as to identify and extract the system fault device sets associated with these false faults. Thereby locating the system devices that may cause false faults in the photovoltaic inverter, providing a basis for further fault troubleshooting and diagnosis.

[0043] Exemplarily, first, invoke the comprehensive system operation and maintenance data of the photovoltaic power generation system. These data usually include but are not limited to: real-time operation parameters of each device, historical operation records, fault alarm information, maintenance records, and environmental parameters (such as temperature, humidity, wind speed, etc.). These data are continuously collected and stored through various methods such as monitoring systems, sensor networks, or data centers, and can provide a rich data source for subsequent fault analysis.

[0044] Next, use the fault time nodes of multiple pseudo-faults obtained in the previous steps as key clues. These time nodes identify possible abnormal or false alarm moments in the system and are of great significance for locating real faults. Use these time nodes as query conditions to traverse and filter the system operation and maintenance data. During the traversal process, use data analysis techniques to identify system anomalies associated with the fault time nodes. These techniques may include time series analysis, association rule mining, clustering analysis, etc. By comparing the data changes before and after the fault time nodes, signs such as abnormal fluctuations in device operation parameters, frequent triggering of fault alarms, and possible mutual influences between devices are found.

[0045] Based on the above analysis, gradually construct multiple system fault device sets. Each device set contains a group of devices that exhibit anomalies at specific fault time nodes. These devices may be the direct cause of the pseudo-fault or associated devices affected by the fault. Record the detailed information of these devices, including device type, location, fault manifestations, and possible fault causes, to obtain multiple system fault device sets. These device sets provide important reference bases for subsequent fault handling, device maintenance, and system optimization.

[0046] P50: Construct a fault troubleshooting model based on the K types of fault types and multiple system fault device sets.

[0047] Further, as Figure 2 shown, step P50 of the embodiment of the present application further includes:

[0048] P51: Acquire multiple device operation data from the device operation and maintenance data based on the multiple fault time nodes; P52: Analyze the multiple device operation data through the inverter fault identification model to classify the multiple pseudo-faults, and obtain K groups of fault time nodes when the K types of fault types are pseudo-faults; P53: Divide the multiple system fault device sets based on the K groups of fault time nodes to obtain K groups of system fault device sets; P54: Use the K types of fault types as topological nodes, and construct K fault troubleshooting networks by using the K groups of system fault device sets; P55: Associatively store the K types of fault types and the K fault troubleshooting networks based on the knowledge graph to obtain the fault troubleshooting model.

[0049] It should be understood that based on the K types of fault types and multiple system fault device sets, a fault troubleshooting model is constructed to achieve comprehensive fault analysis of the photovoltaic inverter and its associated devices.

[0050] First, based on multiple fault time nodes, further collect the operation data of multiple devices from the device operation and maintenance data. These device operation data include key parameters such as voltage, current, and power during the pseudo-fault time nodes, as well as the changes of these parameters in the system. The purpose of this step is to obtain more detailed device behavior data to accurately distinguish different types of faults in subsequent analysis.

[0051] Next, analyze the collected operation data of multiple devices through an inverter fault identification model to classify multiple pseudo-faults. The inverter fault identification model is based on K types of fault types and K fault operation data sets constructed in the early stage, and can effectively identify and classify pseudo-faults to obtain K sets of fault time nodes, which correspond to various operation characteristics of the inverter when pseudo-faults occur.

[0052] Then, take K types of fault types as topological nodes, and construct K fault troubleshooting networks based on the divided K sets of system fault device sets. These networks center around the fault types and organize information such as related devices, fault phenomena, and troubleshooting paths in the form of graphs or networks. Each network represents the troubleshooting logic and steps of a fault type, providing clear fault troubleshooting paths and solutions for operation and maintenance personnel. Through this networked structure, the relationship between fault types and system devices can be more intuitively displayed, facilitating further fault analysis and location.

[0053] Finally, based on knowledge graph technology, associate and store K types of fault types and K fault troubleshooting networks to obtain the final fault troubleshooting model. A knowledge graph is a graph structure used to represent complex association relationships, which can effectively store and manage the association information between fault types and troubleshooting networks. By organizing this information into a knowledge graph, intelligent retrieval and association analysis of the fault troubleshooting model can be realized, thereby improving the accuracy and flexibility of the troubleshooting model. The finally constructed fault troubleshooting model can quickly and accurately perform fault diagnosis and troubleshooting when faults occur in photovoltaic inverters and their associated devices, ensure the reliable operation of the photovoltaic power generation system, and provide effective technical support for operation and maintenance personnel.

[0054] Furthermore, step P54 of the embodiment of the present application further includes:

[0055] P54-1: Invoke the first set of system fault devices from the K sets of system fault devices based on the first fault type, where the first fault type is any one of the K fault types; P54-2: Aggregate the devices in the first set of system fault devices to obtain multiple associated fault devices; P54-3: Use the first fault type and the multiple associated fault devices as topological nodes, connect the topological nodes using the first set of system fault devices, and use the K sets of fault time nodes to identify the topological node connections to obtain the first initial fault relationship network; P54-4: Preset a fault interval threshold and a node connection frequency threshold; P54-5: Correct the first initial fault relationship network using the fault interval threshold and the node connection frequency threshold to obtain the first fault troubleshooting network; P54-6: By analogy, obtain the K fault troubleshooting networks.

[0056] Optionally, the specific process of constructing the K fault troubleshooting networks can be as follows. First, invoke the first set of system fault devices from the K sets of system fault devices based on the first fault type. Here, the first fault type can be any one of the K fault types, and the purpose is to screen out the set of system fault devices associated with a specific fault type. These device sets reflect which devices in the system may be abnormal or faulty when this specific fault type occurs.

[0057] Next, aggregate the devices in the first set of system fault devices, that is, classify and merge these system fault devices according to their functions, locations, or fault characteristics to obtain multiple associated fault devices. These associated fault devices refer to a group of devices that exhibit similar fault characteristics or are affected by the same fault mechanism under a specific fault type.

[0058] Furthermore, use the first fault type and the multiple associated fault devices as topological nodes, and build the connection relationships between these nodes based on the first set of system fault devices. Specifically, use the fault time nodes as the identifiers for the topological node connections, so that the fault associations of different devices at specific time nodes can be clearly represented. In this way, the first initial fault relationship network is formed, which is a topological graph reflecting the relationship between the first fault type and the multiple associated fault devices.

[0059] To further optimize the fault relationship network, two important thresholds are preset: the fault interval threshold and the node connection frequency threshold. The fault interval threshold is used to limit the time interval between different fault events to ensure that only fault events that are temporally close and may be related are connected together; the node connection frequency threshold is used to screen out those device pairs that frequently appear in multiple fault time nodes to exclude accidental associations and enhance the accuracy of fault troubleshooting.

[0060] Next, the first initial fault relationship network is corrected using the preset fault interval threshold and node connection frequency threshold mentioned above. The purpose of the correction is to eliminate misconnections or irrelevant device connections, so as to obtain a more accurate first fault troubleshooting network. This step ensures that each connection in the fault troubleshooting network has practical relevance by strictly screening and adjusting the node connections.

[0061] Finally, following the above steps, the same operations are performed on other fault types by analogy, and finally K fault troubleshooting networks are obtained. Each fault troubleshooting network corresponds to a specific fault type and clearly shows the topological relationship between the fault type and related devices. It can provide an accurate troubleshooting path and diagnostic basis when the photovoltaic inverter fails, helping to quickly locate the fault source and take corresponding maintenance measures to ensure the efficient and stable operation of the photovoltaic power generation system.

[0062] P60: Analyze the real-time operation data obtained by monitoring the photovoltaic inverter through the collaborative operation of the inverter fault identification model and the fault troubleshooting model, so as to perform adaptive fault diagnosis on the photovoltaic inverter.

[0063] Furthermore, step P60 of the embodiment of the present application further includes:

[0064] P61: Synchronize the real-time operation data obtained by monitoring the photovoltaic inverter to the inverter fault identification model, and synchronously run the K fault identification branches in the inverter fault identification model to analyze the real-time operation data for fault diagnosis of the photovoltaic inverter; P62: If the inverter fault identification model diagnoses a real-time fault, synchronize the real-time fault to the fault troubleshooting model for fault type comparison, and output the target fault troubleshooting network; P63: Send the target fault troubleshooting network and the real-time fault as a fault troubleshooting guide to the operation and maintenance center of the photovoltaic power generation system.

[0065] In a possible embodiment of the present application, through the collaborative operation of the fault identification model and the fault troubleshooting model of the photovoltaic inverter, the real-time operation data of the photovoltaic inverter is analyzed to achieve adaptive fault diagnosis.

[0066] First, synchronize the real-time operation data monitored from the photovoltaic inverter to the inverter fault identification model. The real-time operation data includes the latest readings of key parameters such as voltage, current, power, and temperature. The synchronization of this data ensures that the fault identification model can immediately obtain the latest device status information. Next, synchronously operate the K fault identification branches in the fault identification model. Each fault identification branch corresponds to a specific fault type and is analyzed based on the previously constructed fault identification model. These branches compare the real-time operation data with the pre-trained fault features to determine whether the photovoltaic inverter is in a certain fault state. Through this parallel analysis method, the system can quickly detect possible faults in the inverter and improve the response speed of diagnosis.

[0067] If the inverter fault identification model diagnoses a real-time fault, the real-time fault information will be synchronized to the fault troubleshooting model. The fault troubleshooting model will compare the real-time fault with the K fault troubleshooting networks constructed previously to determine the fault type. The comparison process aims to confirm the association between the real-time fault and a specific fault type and locate the possible faulty source device. Through this comparison, accurately determine the target fault troubleshooting network most relevant to the real-time fault. This network contains all system devices related to the fault type and their possible fault states, providing a clear path for further fault handling.

[0068] Furthermore, integrate the target fault troubleshooting network and the real-time fault information into a fault troubleshooting guide. This guide details the fault type, possible faulty devices, and their relationships, providing a clear fault troubleshooting route for the operation and maintenance personnel. Send the fault troubleshooting guide to the operation and maintenance center of the photovoltaic power generation system. The personnel in the operation and maintenance center can quickly locate the faulty device based on the information in the guide and perform corresponding inspection and repair operations. Through this timely feedback mechanism, the system can significantly improve the efficiency of fault handling, reduce the downtime of the photovoltaic power generation system, reduce losses caused by misjudgment or missed judgment, and ensure the continuous and stable operation of the system.

[0069] In summary, the embodiments of the present application at least have the following technical effects:

[0070] The present application calls the device operation and maintenance data of the photovoltaic inverter, constructs an inverter fault identification model, traverses the system operation and maintenance data for data collection based on multiple fault time nodes of multiple false faults, constructs and generates a fault troubleshooting model, and finally collaborates with the inverter fault identification model and the fault troubleshooting model to perform real-time operation data analysis of the photovoltaic inverter for adaptive fault diagnosis.

[0071] It achieves the technical effect of effectively distinguishing true faults from false faults of the photovoltaic inverter, reducing false alarms and false shutdowns, and improving the overall reliability of the photovoltaic system.

[0072] Embodiment 2, based on the same inventive concept as the adaptive fault diagnosis method for a photovoltaic inverter in the foregoing embodiment, as Figure 3 shown, the present application provides an adaptive fault diagnosis device for a photovoltaic inverter. The device in the embodiment of the present application and the method embodiment are based on the same inventive concept. Among them, the device includes:

[0073] A fault operation data extraction module 11, which is configured to call the device operation and maintenance data of the photovoltaic inverter and extract K fault operation data sets of K fault types from the device operation and maintenance data based on the fault type.

[0074] An inverter fault identification model construction module 12, which is configured to construct and generate an inverter fault identification model based on the K fault types and the K fault operation data sets.

[0075] A fault time node acquisition module 13, which is configured to collect multiple fault time nodes of multiple false faults from the device operation and maintenance data.

[0076] A system fault device collection module 14, which is configured to call the system operation and maintenance data of the photovoltaic power generation system and traverse the system operation and maintenance data by using the multiple fault time nodes to obtain multiple system fault device sets.

[0077] A fault troubleshooting model construction module 15, which is configured to construct and generate a fault troubleshooting model based on the K fault types and the multiple system fault device sets.

[0078] An adaptive fault diagnosis module 16, which is configured to perform data analysis on the real-time operation data obtained by monitoring the photovoltaic inverter through the cooperative operation of the inverter fault identification model and the fault troubleshooting model, so as to perform adaptive fault diagnosis on the photovoltaic inverter.

[0079] Further, the inverter fault identification model construction module 12 is further configured to perform the following steps:

[0080] Preset the sample size requirement. When any one of the K faulty operation datasets does not meet the sample size requirement, interactively obtain the photovoltaic system attributes of the photovoltaic power generation system; based on the photovoltaic system attributes, call and obtain the M sample operation and maintenance data of M sample inverters from a pre-constructed power generation system information library, where M is a positive integer; use the M sample operation and maintenance data to update the K faulty operation datasets to obtain K updated operation datasets; construct a standard fault identification model based on the gradient boosting machine algorithm; use the K updated operation datasets as training data for training and tuning the standard fault identification model to obtain K fault identification branches; parallelize the K fault identification branches to complete the construction of the inverter fault identification model. The photovoltaic system attributes include the target installation capacity, target conversion efficiency, target system efficiency, target performance ratio, target sunshine hours, and target radiation level.

[0081] Further, the inverter fault identification model construction module 12 is further configured to perform the following steps:

[0082] Interactively obtain multiple sample power generation systems with the same photovoltaic panel type as the photovoltaic power generation system, and call the multiple sample system attributes and multiple sample operation and maintenance data of the multiple sample power generation systems; based on the knowledge graph, associate and store the multiple sample power generation systems, multiple sample system attributes, and multiple sample operation and maintenance data to obtain the power generation system information library; calculate multiple attribute similarity coefficients between the photovoltaic system attributes and multiple sample system attributes based on the Euclidean distance and custom weighted distance; preset an attribute similarity threshold, and call and obtain the M sample operation and maintenance data of M sample power generation systems whose attribute similarity coefficients meet the attribute similarity threshold from the power generation system information library, where the M sample operation and maintenance data are data records of the operation and maintenance of the M sample inverters in the M sample power generation systems.

[0083] Further, the fault troubleshooting model construction module 15 is further configured to perform the following steps:

[0084] Collect multiple device operation data from the device operation and maintenance data based on the multiple fault time nodes; analyze the multiple device operation data through the inverter fault identification model to classify the multiple false faults, and obtain K groups of fault time nodes when the K fault types are false faults; divide the multiple system fault device sets based on the K groups of fault time nodes to obtain K groups of system fault device sets; use the K fault types as topological nodes, and construct K fault troubleshooting networks by using the K groups of system fault device sets; based on the knowledge graph, associate and store the K fault types and K fault troubleshooting networks to obtain the fault troubleshooting model.

[0085] Further, the fault troubleshooting model construction module 15 is further configured to perform the following steps:

[0086] Synchronize the real-time operation data obtained by monitoring the PV inverter to the inverter fault identification model, and synchronously run the K fault identification branches in the inverter fault identification model to analyze the real-time operation data for fault diagnosis of the PV inverter; if the inverter fault identification model diagnoses a real-time fault, synchronize the real-time fault to the fault troubleshooting model for fault type comparison, and output a target fault troubleshooting network; send the target fault troubleshooting network and the real-time fault to the operation and maintenance center of the PV power generation system as a fault troubleshooting guide.

[0087] Further, the adaptive fault diagnosis module 16 is further configured to perform the following steps:

[0088] Call the first set of system fault devices from the K sets of system fault device sets based on the first fault type, where the first fault type is any one of the K fault types; perform device aggregation on the first set of system fault devices to obtain multiple associated fault devices; use the first fault type and the multiple associated fault devices as topological nodes, connect the topological nodes using the first set of system fault devices, and use the K sets of fault time nodes to identify the topological node connections to obtain a first initial fault relationship network; preset a fault interval threshold and a node connection frequency threshold; correct the first initial fault relationship network using the fault interval threshold and the node connection frequency threshold to obtain a first fault troubleshooting network; and so on to obtain the K fault troubleshooting networks.

[0089] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above description of specific embodiments of this specification is given. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0090] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0091] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An adaptive fault diagnosis method for a photovoltaic inverter, characterized in that: The method comprises: Calling equipment operation and maintenance data of the photovoltaic inverter, and extracting K fault operation data sets of K fault types from the equipment operation and maintenance data based on the fault type; Constructing and generating an inverter fault identification model based on the K fault types and the K fault operation data sets; Acquire multiple fault time nodes of multiple false faults from the equipment operation and maintenance data; Calling system operation and maintenance data of the photovoltaic power generation system, and traversing the system operation and maintenance data using the multiple fault time nodes to obtain multiple system fault device sets; Building a fault troubleshooting model based on the K fault types and multiple system fault device sets; The inverter fault identification model and the fault troubleshooting model operating in coordination perform data analysis on real-time operating data obtained by monitoring the photovoltaic inverter, so as to perform adaptive fault diagnosis on the photovoltaic inverter; Constructing and generating an inverter fault identification model based on the K fault types and the K fault operation data sets includes: Preset a sample size requirement, and when any fault operation data set among the K fault operation data sets does not meet the sample size requirement, interactively obtain photovoltaic system properties of the photovoltaic power generation system; Based on the photovoltaic system attributes, M sample operation and maintenance data of M sample inverters are obtained from a pre-built power generation system information library, where M is a positive integer; Using the M sample operation and maintenance data to update the K fault operation data sets to obtain K updated operation data sets; Construct a standard fault identification model based on the gradient boosting machine algorithm; Using the K updated running data sets as training data to perform training and parameter adjustment on the standard fault recognition model, to obtain K fault recognition branches; Connecting the K fault identification branches in parallel to complete the construction of the inverter fault identification model; A fault troubleshooting model is constructed based on the K fault types and multiple system fault device sets, including: Acquire multiple equipment operation data from the equipment operation and maintenance data based on the multiple fault time nodes; Analyzing the plurality of equipment operation data through the transformer fault identification model to classify the plurality of false faults, and obtaining K groups of fault time nodes when the K types of faults are false faults; Divide the multiple system fault device sets based on the K groups of fault time nodes to obtain K groups of system fault device sets; Taking the K types of faults as topological nodes, and using the K groups of system fault device sets to construct K fault troubleshooting networks; Based on the knowledge graph, the K fault types and the K fault troubleshooting networks are stored in association to obtain the fault troubleshooting model; The K types of faults are used as topological nodes, and the K groups of system fault device sets are used to construct K fault troubleshooting networks, including: Calling a first set of system fault device sets from the K sets of system fault device sets based on a first fault type, wherein the first fault type is any one of the K fault types; Aggregating the first group of system faulty devices to obtain multiple associated faulty devices; The first fault type and multiple associated fault devices are used as topological nodes, the first group of system fault device sets are used to connect the topological nodes, and the K groups of fault time nodes are used to identify the topological node connections to obtain a first initial fault relationship network; Preset fault interval threshold and node connection frequency threshold; Correcting the first initial fault relationship network using the fault interval threshold and the node connection frequency threshold to obtain a first fault troubleshooting network; And so on, the K fault troubleshooting networks are obtained.

2. The method for adaptive fault diagnosis of a photovoltaic inverter according to claim 1, characterized in that: Based on the photovoltaic system attributes, M sample operation and maintenance data of M sample inverters are obtained from a pre-built power generation system information library, and the method further includes: Interactively obtain multiple sample power generation systems of the same photovoltaic panel type as the photovoltaic power generation system, and call multiple sample system attributes and multiple sample operation and maintenance data of the multiple sample power generation systems; The plurality of sample power generation systems, the plurality of sample system attributes and the plurality of sample operation and maintenance data are stored in association with each other based on the knowledge graph to obtain the power generation system information library; Based on the Euclidean distance and the custom weighted distance, multiple attribute similarity coefficients between the photovoltaic system attribute and multiple sample system attributes are calculated; A property similarity threshold is preset, and the M sample operation and maintenance data of the M sample power generation systems whose attribute similarity coefficients meet the property similarity threshold are obtained from the power generation system information library, wherein the M sample operation and maintenance data are data records for operation and maintenance of the M sample inverters in the M sample power generation systems.

3. The method for adaptive fault diagnosis of a photovoltaic inverter according to claim 2, characterized in that: The photovoltaic system attributes include target installation capacity, target conversion efficiency, target system efficiency, target performance ratio, target sunshine hours and target radiation level.

4. The method for adaptive fault diagnosis of a photovoltaic inverter according to claim 1, characterized in that: The inverter fault identification model and the fault troubleshooting model operating in coordination perform data analysis on real-time operation data obtained by monitoring the photovoltaic inverter to perform adaptive fault diagnosis on the photovoltaic inverter. The method further includes: Synchronizing the real-time operation data obtained by monitoring the photovoltaic inverter to the inverter fault identification model, and synchronously running the K fault identification branches in the inverter fault identification model to analyze the real-time operation data to perform fault diagnosis of the photovoltaic inverter; If the inverter fault identification model diagnoses a real-time fault, the real-time fault is synchronized to the fault troubleshooting model for fault type comparison, and a target fault troubleshooting network is output; The target fault troubleshooting network and the real-time fault are sent as a fault troubleshooting guide to an operation and maintenance center of the photovoltaic power generation system.

5. An adaptive fault diagnosis device for a photovoltaic inverter, characterized in that: The device is used to execute an adaptive fault diagnosis method for a photovoltaic inverter according to any one of claims 1 to 4, and the device comprises: A fault operation data extraction module, the fault operation data extraction module is used to call the equipment operation and maintenance data of the photovoltaic inverter, and extract K fault operation data sets of K fault types from the equipment operation and maintenance data based on the fault type; An inverter fault identification model construction module, the inverter fault identification model construction module is used to construct and generate an inverter fault identification model based on the K fault types and the K fault operation data sets; A fault time node acquisition module, the fault time node acquisition module is used to acquire multiple fault time nodes of multiple false faults from the equipment operation and maintenance data collection; A system fault equipment collection module, the system fault equipment collection module is used to call the system operation and maintenance data of the photovoltaic power generation system, and use the multiple fault time nodes to traverse the system operation and maintenance data to obtain multiple system fault equipment sets; A fault troubleshooting model building module, the fault troubleshooting model building module is used to build and generate a fault troubleshooting model based on the K types of fault types and multiple system fault device sets; An adaptive fault diagnosis module, wherein the adaptive fault diagnosis module is used to analyze the real-time operation data obtained by monitoring the photovoltaic inverter by using the inverter fault identification model and the fault troubleshooting model in coordination, so as to perform adaptive fault diagnosis on the photovoltaic inverter; The inverter fault identification model building module is used to interactively obtain multiple sample power generation systems of the same photovoltaic panel type as the photovoltaic power generation system, and call multiple sample system attributes and multiple sample operation and maintenance data of the multiple sample power generation systems; based on the knowledge graph, the multiple sample power generation systems, multiple sample system attributes and multiple sample operation and maintenance data are associated and stored to obtain the power generation system information library; based on the Euclidean distance and the custom weighted distance, multiple attribute similarity coefficients between the photovoltaic system attributes and the multiple sample system attributes are calculated; a preset attribute similarity threshold is preset, and the M sample operation and maintenance data of the M sample power generation systems whose attribute similarity coefficients meet the attribute similarity threshold are called from the power generation system information library to obtain, wherein the M sample operation and maintenance data are data records for operation and maintenance of the M sample inverters in the M sample power generation systems.

Images