Power transformation equipment fault diagnosis method based on inspection thinking chain and related device

Abstract

The invention discloses a power transformation equipment fault diagnosis method based on an inspection thinking chain and a related device, and relates to the technical field of fault diagnosis, and the method comprises the steps: obtaining standing book information and an abnormal phenomenon of to-be-detected equipment; inputting the standing book information of the to-be-tested equipment and the abnormal phenomenon into a fault diagnosis model to obtain multi-round dialogue type fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained based on large model training; the multi-round dialogue type fault diagnosis feedback comprises a fault reason checking process; according to the multi-round dialogue type fault diagnosis feedback, a current checking phenomenon is input step by step, and fault reason troubleshooting is carried out; after the troubleshooting of the fault reason is finished, generating a fault diagnosis result; the fault diagnosis result comprises the follow-up processing condition and rectification measures of the fault. According to the invention, the processing capability of a large model on complex fault diagnosis scene problems can be improved; and the availability of the large model application is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method for diagnosing faults of substation equipment based on an inspection thinking chain and related devices. Background Art

[0002] In recent years, the number of power equipment and facilities in new power systems has exploded, with a large number of heterogeneous terminals connected to the grid. This has posed significant challenges to the observability, measurability, controllability, and cybersecurity of power systems. Power equipment is a complex and critical component of the power system, and failures in different locations and at different types can directly impact the efficient operation of the power system. Among these, power transformers, high-voltage circuit breakers, and gas-insulated metal-enclosed combination units (GIS) are key substation equipment in the power grid. Accurately assessing the operating status of these substations and implementing timely, effective operation and maintenance strategies are crucial to ensuring reliable grid operation and national energy security.

[0003] The operation and maintenance of power system equipment can usually be divided into the following three types: one is passive maintenance, also known as accident maintenance or post-maintenance, which mainly takes corresponding solutions to power system equipment failure problems; the second is planned maintenance. This type of maintenance work is mainly preventive. Electrical equipment maintenance personnel generally conduct regular inspections after the daily operation of the equipment. Once problems are found in the equipment, they will carry out repairs in time. At this stage of work, a lot of manpower and material resources are often invested; the third is condition maintenance. The maintenance work at this stage needs to be closely integrated with the specific operating conditions of the equipment, and advanced equipment detection and diagnosis methods are used to promptly detect potential equipment failures or safety hazards.

[0004] From a current perspective, my country's power companies primarily use regular maintenance to troubleshoot power equipment during operation. However, due to the complex structure and volatile working environment of substation equipment, coupled with the continuous expansion of power grid construction, the increasing number of equipment, the increasingly complex grid structure, and the shortage of staff, power companies currently struggle to ensure the orderly, efficient, and high-quality implementation of grid maintenance. Furthermore, power companies accumulate large amounts of long text data during equipment maintenance, including equipment failure reports, test and inspection reports, and standard guidelines. While this text data has been documented, the knowledge value contained in this historical data has not been fully explored. Therefore, improving the processing capabilities of large models for complex fault diagnosis scenarios and enhancing the usability of large model applications have become urgent technical challenges in this field. Summary of the Invention

[0005] The purpose of the present invention is to provide a substation fault diagnosis method and related devices based on the inspection thinking chain, which can improve the processing ability of large models in complex fault diagnosis scenarios and improve the availability of large model applications.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] In a first aspect, the present invention provides a method for diagnosing a fault of a substation equipment based on an inspection thinking chain, the method comprising:

[0008] Obtain the ledger information and abnormal phenomena of the device under test.

[0009] The inventory information of the device under test and the abnormal phenomenon are input into the fault diagnosis model to obtain multi-round conversational fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained based on large model training; the multi-round conversational fault diagnosis feedback includes the process of troubleshooting the cause of the fault.

[0010] Based on the multiple rounds of dialogue-based fault diagnosis feedback, the current inspection phenomenon is gradually input to conduct fault cause investigation.

[0011] After the cause of the fault is checked, a fault diagnosis result is generated; the fault diagnosis result includes: the subsequent handling of the fault and corrective measures.

[0012] In a second aspect, the present invention provides a device for diagnosing faults of substation equipment based on an inspection thinking chain, the device for diagnosing faults of substation equipment based on an inspection thinking chain comprising:

[0013] The data acquisition module is used to obtain the ledger information and abnormal phenomena of the equipment under test.

[0014] An input module is used to input the inventory information of the device under test and the abnormal phenomenon into a fault diagnosis model to obtain multi-round conversational fault diagnosis feedback from the fault diagnosis model; the fault diagnosis model is a model obtained through training based on a large model; the multi-round conversational fault diagnosis feedback includes the process of troubleshooting the cause of the fault.

[0015] The fault cause investigation module is used to gradually input the current inspection phenomenon based on the multiple rounds of dialogue-based fault diagnosis feedback to conduct fault cause investigation.

[0016] The fault diagnosis result generating module is used to generate the fault diagnosis result after the fault cause is checked. The fault diagnosis result includes the subsequent handling of the fault and the corrective measures.

[0017] In a third aspect, the present invention provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any one of the above-mentioned methods for diagnosing faults of substation equipment based on the inspection thinking chain.

[0018] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any one of the above-mentioned methods for diagnosing faults of substation equipment based on the inspection thinking chain.

[0019] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which, when executed by a processor, implements any one of the above-mentioned methods for diagnosing faults of substation equipment based on the inspection thinking chain.

[0020] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0021] The present invention provides a method for diagnosing faults of substation equipment based on an inspection thinking chain and related devices, the method comprising: obtaining the ledger information and abnormal phenomena of the equipment to be tested; inputting the ledger information and the abnormal phenomena of the equipment to be tested into a fault diagnosis model to obtain multi-round dialogue-style fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained based on training of a large model; the multi-round dialogue-style fault diagnosis feedback includes the process of troubleshooting the cause of the fault; based on the multi-round dialogue-style fault diagnosis feedback, the current inspection phenomenon is gradually input to troubleshoot the cause of the fault; when the troubleshooting of the cause of the fault is completed, a fault diagnosis result is generated; the fault diagnosis result includes: the subsequent handling of the fault and corrective measures. The present invention improves the processing capability of the large model in complex fault diagnosis scenarios by guiding the fault diagnosis process; using a multi-round dialogue method to interact with the user can more accurately understand the fault phenomenon, collect necessary information, and adjust the diagnosis direction according to the user's feedback, thereby improving the usability of the large model application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 This is an application environment diagram of a method for diagnosing faults of substation equipment based on an inspection thinking chain in one embodiment of the present invention.

[0024] Figure 2 A flowchart of a method for diagnosing faults of substation equipment based on an inspection thinking chain is provided in accordance with an embodiment of the present invention.

[0025] Figure 3 A schematic diagram of the overall process provided by an embodiment of the present invention.

[0026] Figure 4 A schematic diagram of functional modules of a device for diagnosing faults of substation equipment based on an inspection thinking chain is provided in accordance with an embodiment of the present invention.

[0027] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0029] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] For power equipment, the "double-high" characteristics of the current new power system have changed the operating characteristics of traditional equipment, necessitating further improvements in equipment intelligence and enhanced comprehensive sensing capabilities. Furthermore, the new power system places new demands on the digitalization of equipment, necessitating the in-depth application of artificial intelligence technology and the advancement of digital applications in diagnosis, operation, and maintenance.

[0031] Artificial intelligence technology is a major future development direction. Against the backdrop of rapidly growing data and the continuous development of computing resources, AI big model technologies, represented by ChatGPT, have achieved rapid breakthroughs and have demonstrated significant application value across multiple domains, scenarios, and tasks. Deeper vertical applications in the energy and power sectors are also showing urgent demand and promising prospects. With the continuous development of smart grids, information systems such as equipment management systems (PMS), energy management systems (EMS), and information management systems (MIS) serving various disciplines have been gradually built and implemented, and the scale of grid data continues to expand. This data, including sensor data related to equipment status, historical maintenance records, and real-time monitoring images, has gradually become characterized by large volume, diverse types, and rapid growth. This data foundation has laid a data foundation for the application and development of big model technology in power equipment condition assessment and intelligent operation and maintenance. Using big models, combined with machine learning and deep learning technologies, can enable real-time monitoring and predictive analysis of the operating status of power equipment, promptly identifying potential faults and anomalies and providing early warnings. By analyzing historical and real-time data, big models can provide scientific decision support for maintenance teams, optimize maintenance plans and resource allocation, and reduce unnecessary maintenance costs.

[0032] While large models demonstrate powerful capabilities when processing general-purpose data and tasks, they face challenges and limitations in specific verticals, such as the power and healthcare sectors. This includes the occurrence of "hallucinations," whereby answers are fabricated without knowing the answer, sometimes even using fictitious references. These hallucinations occur because large models are typically trained on general-purpose datasets and lack specialized domain expertise and in-depth understanding of specific fields.

[0033] To address the "hallucination" problem of large models in specific domains, we need to infuse them with specific domain knowledge through fine-tuning. This involves retraining the large model to enable it to understand and adapt to data and tasks in a specific domain. Fine-tuning is a technique that retrains an already trained model to help it adapt to new data and tasks. During fine-tuning, we use datasets from specific domains to train the model, enabling it to learn the domain expertise and patterns. However, the diversity and complexity of power data makes it difficult to construct a fine-tuning dataset for question-answering.

[0034] Therefore, in order to solve the above problems, the present invention proposes a substation fault diagnosis method and related devices based on the inspection thinking chain. This method gradually simulates the troubleshooting and repair process of operation and maintenance experts when facing specific faults by constructing a maintenance thinking chain, which includes starting from the fault phenomenon and gradually analyzing the possible causes. And through function calls, it integrates various professional tools in the power field involved in the maintenance process, such as waveform comparison tools, oil chromatography analysis tools, etc., to assist in complex fault diagnosis. Interacting with users in the form of multiple rounds of dialogue can more accurately understand the fault phenomenon, collect necessary information, and adjust the diagnosis direction according to user feedback.

[0035] The power substation fault diagnosis method based on the inspection thinking chain provided by the embodiment of the present invention can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the ledger information and abnormal phenomena of the device under test to the server 104. After the server 104 receives the ledger information and abnormal phenomena of the device under test, the server 104 inputs the ledger information and abnormal phenomena of the device under test into the fault diagnosis model to obtain multi-round dialogue-style fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained by training based on a large model; the multi-round dialogue-style fault diagnosis feedback includes the process of troubleshooting the cause of the fault; according to the multi-round dialogue-style fault diagnosis feedback, the current inspection phenomenon is gradually input to troubleshoot the cause of the fault; when the cause of the fault is troubleshooted, a fault diagnosis result is generated; the fault diagnosis result includes: the subsequent handling of the fault and the corrective measures. The server 104 can feedback the obtained fault diagnosis result to the terminal 102. In addition, in some embodiments, the substation equipment fault diagnosis method based on the inspection thinking chain can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly perform fault diagnosis on the ledger information and abnormal phenomena of the equipment to be tested, or the server 104 can obtain the ledger information and abnormal phenomena of the equipment to be tested from the data storage system, and perform fault diagnosis on the ledger information and abnormal phenomena of the equipment to be tested.

[0036] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, and tablet computers. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or a cloud server.

[0037] In an exemplary embodiment, Figure 2 As shown, a method for fault diagnosis of substation equipment based on inspection thinking chain is provided. The method is executed by a computer device, specifically a computer device such as a terminal or a server, or a terminal and a server. In an embodiment of the present invention, the method is applied to Figure 1 The server 104 in the example is used for explanation, and the steps include the following steps S1 to S4.

[0038] in:

[0039] S1: Obtain the record information and abnormal phenomena of the device under test.

[0040] S2: Input the inventory information of the device under test and the abnormal phenomenon into the fault diagnosis model to obtain multi-round conversational fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained by training based on a large model; the multi-round conversational fault diagnosis feedback includes the process of troubleshooting the cause of the fault.

[0041] S3: Based on the multiple rounds of dialogue-based fault diagnosis feedback, the current inspection phenomenon is gradually input to troubleshoot the cause of the fault.

[0042] S4: After the fault cause is checked, a fault diagnosis result is generated; the fault diagnosis result includes: the subsequent handling of the fault and the corrective measures.

[0043] By implementing the above-mentioned steps S1 to S4, the present invention improves the processing capability of the large model in complex fault diagnosis scenarios by guiding the fault diagnosis process; using a multi-round dialogue method to interact with the user can more accurately understand the fault phenomenon, collect necessary information, and adjust the diagnosis direction according to the user's feedback, thereby improving the usability of the large model application.

[0044] In an exemplary embodiment, in step S2, the fault diagnosis model is determined as follows:

[0045] S21: Based on historical fault repair reports, determine the inspection thinking chain and function registration data for different faults; the inspection thinking chain includes: equipment ledger information, fault phenomenon information, fault inspection process and fault handling; the function registration data is the corresponding external function or service called according to the instruction, which is used to integrate various professional tools involved in the maintenance process.

[0046] S22: Construct a fine-tuning data set based on the inspection thinking chain and the function registration data; the fine-tuning data set includes: fault ledger information, fault abnormality, fault cause investigation process, subsequent processing status and rectification measures.

[0047] S23: Label the fine-tuning dataset to obtain a labeled multi-round conversational fine-tuning dataset.

[0048] S24: Using the fault log information and fault anomalies in the labeled multi-round conversational fine-tuning dataset as input, and the fault cause investigation process, subsequent processing status, and corrective measures corresponding to the fault log information and fault anomalies as output, the large model is trained through multi-round conversational fine-tuning to obtain a fault diagnosis model.

[0049] In step S21, the construction of the inspection thinking chain is as follows:

[0050] Because different fault phenomena require different inspection methods and steps, the method of the present invention primarily extracts inspection chains for different faults from data such as substation fault reports. Based on the content of the substation fault report, a step-by-step thinking chain is constructed. Each node in the constructed chain corresponds to an inspection step, and the connections between the nodes represent the logical relationship between the inspection steps.

[0051] A specific thinking chain includes four parts: equipment inventory information, fault phenomenon information, fault inspection process and fault handling.

[0052] Equipment inventory information is the basic information of the faulty equipment, including equipment type, equipment model, voltage level, commissioning time, etc.

[0053] Fault phenomenon information refers to abnormal equipment phenomena observed during equipment alarms or inspections. These abnormal phenomena may include, but are not limited to, unusual sounds, vibrations, elevated temperatures, unusual gases, flashing or off indicators, and abnormal instrument readings. These phenomena serve as the starting point for fault diagnosis and provide preliminary guidance for further investigation.

[0054] The fault diagnosis process involves a series of detailed inspections of the faulty device based on the fault symptom information. This may include an external inspection of the device to confirm whether there is any obvious damage or wear; electrical tests such as insulation resistance and ground resistance tests to evaluate the device's electrical performance; and software diagnostics to verify that the device's control logic and programming are functioning properly.

[0055] The fault inspection process is a multi-step process of variable length. Each step is formatted as ['Inspection Type', 'Inspection Finding', 'Inspection Verdict']. The inspection type is the inspection method, such as infrared temperature measurement, visible light damage identification, or live detection. The inspection finding is the findings of the inspection method. The inspection verdict is the judgment after the inspection, such as whether to proceed to the next step of inspection.

[0056] Troubleshooting is a series of specific actions and measures taken in response to identified equipment failures. The goal is to eliminate the failure and restore normal equipment operation. Repair methods may include replacing damaged parts, adjusting equipment parameters, and repairing electrical circuits.

[0057] The specific structure of a complete inspection thinking chain is:

[0058]

[0059]

[0060] The content following the / / symbol is a comment description, and the content in the square brackets <> is user input or system output.

[0061] The process of determining the function registration data is as follows:

[0062] S211: Using the large model, based on the user input information, determine whether a function needs to be called and obtain a determination result.

[0063] S212: If the judgment result is yes, determine the function to be called and the corresponding parameters.

[0064] S213: Generate a structured output according to the function to be called and the corresponding parameters; the structured output includes the parameters required for function calling (Function Calling).

[0065] S214: Calling a corresponding function according to the structured output.

[0066] S215: Obtain function registration data according to the corresponding function.

[0067] Specifically, the function call registration is as follows:

[0068] Function calls are an advanced feature of large language models (such as GPT). They enable the model to determine whether to call corresponding external functions or services based on user instructions and return information in a structured format, rather than simply providing plain text responses. This enhances the model's capabilities and provides richer and more accurate responses. This step registers external tool functions with the large model. Through function calls, various professional tools in the power field involved in the maintenance process, such as waveform comparison tools and oil chromatography analysis tools, are integrated to assist in complex fault diagnosis.

[0069] Describe external utility functions. This requires defining one or more external utility functions according to a specific format. These functions can be external API calls, database queries, or any other code snippets that perform specific tasks. These functions must have clear names and parameter lists.

[0070] Large model parsing. After receiving user input, the large model parses and understands the input. Based on its training data and algorithm, the model determines whether a function needs to be called and determines the function to be called and its parameters.

[0071] Generate a function call. If the model determines that a function needs to be called, it generates a structured output containing the parameters required for the function call, typically a JSON object containing information such as the function name and parameter list.

[0072] Function call execution. The system parses the JSON object generated by the model, converts it into a valid data structure, and calls the corresponding function with these parameters. This process is completed in the code environment of the entire system, rather than within the large model.

[0073] Processing function results. After the function call completes, the results need to be returned to the model. This is typically accomplished by appending the results to the model and calling it again. The model receives and processes these results, then generates a natural language response to the user, summarizing or explaining the function call's results.

[0074] In this way, function calls extend the capabilities of large models, enabling them to perform more complex tasks and interact more richly with the outside world.

[0075] The specific format of external function registration is:

[0076]

[0077]

[0078] In another exemplary embodiment of the present invention, in step S23, the following steps are specifically included:

[0079] S231: Construct prompt words.

[0080] S232: manually filling the prompt words with the fine-tuning dataset to obtain a labeled multi-round conversational fine-tuning dataset.

[0081] Specifically, in order to inject diagnostic knowledge in the field of substation equipment into the large model, it is necessary to form corresponding inspection thinking chains and function registration data for the specific fault diagnosis process, and conduct fine-tuning training.

[0082] The data source is mainly the historical fault maintenance reports of substation equipment. Such reports have a certain writing format and are generally divided into five parts. The first part is the fault ledger information, the second part is the abnormal fault phenomenon, the third part is the fault cause investigation process, the fourth part is the subsequent processing situation, and the fifth part is the corrective measures.

[0083] Generally, manual labeling is used to organize these fault repair reports into corresponding data formats, but this method is labor-intensive. The present invention proposes a semi-automatic method to construct data. The method is to build a detailed prompt word, input it into an open source big model, use the big model to perform preliminary field extraction, and then manually fill the corresponding field content into the corresponding fine-tuning data format.

[0084] The format of the constructed prompt word is:

[0085] Prompt word = "The following is a power equipment failure operation and maintenance report:

[0086] ```

[0087] The text of a fault repair report

[0088] ```

[0089] Please identify the relevant information in the fault report. The following describes each field that needs to be extracted and identified and the corresponding field description. Please output the results in JSON format:

[0090] Substation: The substation where the transformer is located

[0091] Installation location: If the high-end Y / DA phase of the pole II (if any)

[0092] Voltage levels: ±200kV, ±400kV, ±600kV, ±800kV, ±1100kV, etc.

[0093] Model: such as ZZDFPZ-382000 / 500-600, etc.

[0094] Equipment manufacturers: Xi'an Transformer Group, TBEA, Shenyang Transformer Group, Baoding Transformer Group, etc.

[0095] Commissioning time: Equipment commissioning time

[0096] Failure occurrence stage: manufacturing, installation, operation

[0097] Failure (defect) time: Failure time

[0098] Fault (defect) parts: such as oil tank, winding, bushing, tap changer and cooler, etc.

[0099] Fault (defect) phenomena: mainly (equipment tripping, abnormal oil chromatography, abnormal partial discharge, abnormal heating, abnormal tap changer and oil / gas leakage, etc.

[0100] Fault process description: Use text to describe the fault discovery and on-site handling.

[0101] Post-fault inspection: What inspections were performed after the fault? What steps were taken? What issues were addressed?

[0102] Overall conclusion of post-fault inspection:

[0103] Cause of failure (defect): The expression format is "root cause + cause + direct cause", such as insulation discharge caused by quality defects of insulation materials, and suspended discharge caused by loose bolts due to poor assembly process control.

[0104] Troubleshooting steps: Detailed troubleshooting steps.

[0105] Measures and suggestions: Fault operation and maintenance suggestions.

[0106] Please output the results in JSON format, with each field as a key:

[0107] ”.

[0108] Next, we use the constructed fine-tuning dataset in the multi-round dialogue format to fine-tune the large model. Fine-tuning refers to further training based on the pre-trained large language model using a specific dataset to adapt the model to specific tasks or fields, complete knowledge injection and instruction alignment, and verify the model's understanding and application of fault diagnosis logic.

[0109] Fine-tune the large model using a dataset in a multi-round dialogue format. The fine-tuning training loss function uses the cross-entropy loss function for autoregressive prediction. Different loss function masks (loss_mask) are added for different roles, allowing the loss calculation for multiple rounds of large model responses in a single pass. The system and user roles represent inputs and do not participate in model parameter updates, so no corresponding losses need to be calculated. The assistant role represents the large model's responses, for which the corresponding generative cross-entropy loss needs to be calculated.

[0110] LoRA (Low-Rank Adaptation) is a fine-tuning method that updates parameters by introducing a low-rank structure into the model's weight matrix, thereby reducing the number of parameters that need to be updated during training and improving training efficiency. The present invention uses the LoRA fine-tuning algorithm to train large model base models (such as ChatGLM3-7b, etc.) on graphics card hardware environments such as NVIDIA's Tesla to accelerate the training process. At the same time, ensure that there is enough memory and storage space to handle large models and data sets. Training setting parameters such as learning rate, batch size, number of training rounds, LoRA rank (usually low such as 2 or 4), weight decay, etc., use the default settings for large models.

[0111] After fine-tuning and training, the large model is infused with domain knowledge related to substation equipment. Loading this model using the Python Transformer package enables multi-round conversational fault diagnosis services. The user first enters the faulty equipment inventory and basic fault information. The system's fault diagnosis module then uses external functions to diagnose the substation fault based on the received function call parameters and the current node's inspection indicators. If the diagnosis result does not meet the preset threshold, the system proceeds to the next inspection step; otherwise, the system terminates the diagnostic process and outputs the fault diagnosis result.

[0112] In addition, the system will generate a fault diagnosis report based on the fault diagnosis result, and output the fault diagnosis report to the user or expert through the user interface or interface.

[0113] At this point, the entire process of this method is completed. The overall process diagram is as follows Figure 3 shown.

[0114] Compared with the existing fault diagnosis system for substation equipment, the present invention improves the processing ability of large models in complex fault diagnosis scenarios by constructing an inspection thinking chain to guide the fault diagnosis process; and uses function calls to call external diagnostic tools, thereby expanding the diagnostic capabilities of large models; and uses multi-round dialogues to realize interaction with users or experts, thereby improving the usability of large model applications.

[0115] The present invention also provides an application scenario, which applies the above-mentioned substation fault diagnosis method based on the inspection thinking chain. Specifically: the substation fault diagnosis method based on the inspection thinking chain provided in this embodiment can be applied in the fault diagnosis scenario. The fault diagnosis scenario includes: a data acquisition link, a model training link and a fault diagnosis link; the acquired ledger information and abnormal phenomena of the equipment to be tested are input into the trained fault diagnosis model to obtain multi-round conversational fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained based on the large model training; the multi-round conversational fault diagnosis feedback includes the fault cause investigation process; according to the multi-round conversational fault diagnosis feedback, the current inspection phenomenon is gradually input to investigate the cause of the fault; when the fault cause investigation is completed, the fault diagnosis result can be generated; the fault diagnosis result includes: the subsequent processing of the fault and the rectification measures.

[0116] Based on the same inventive concept, embodiments of the present invention further provide an apparatus for implementing the aforementioned method for diagnosing substation faults based on an inspection chain of thought. The solution provided by this apparatus is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more embodiments of the apparatus for diagnosing substation faults based on an inspection chain of thought provided below can be found in the aforementioned limitations of the method for diagnosing substation faults based on an inspection chain of thought, and will not be further elaborated here.

[0117] In an exemplary embodiment, Figure 4 As shown, a substation fault diagnosis device based on an inspection thinking chain is provided, comprising:

[0118] The data acquisition module M1 is used to obtain the ledger information and abnormal phenomena of the device under test.

[0119] Input module M2 is used to input the inventory information of the equipment under test and the abnormal phenomenon into the fault diagnosis model to obtain multi-round conversational fault diagnosis feedback of the fault diagnosis model; the fault diagnosis model is a model obtained based on large model training; the multi-round conversational fault diagnosis feedback includes the process of troubleshooting the cause of the fault.

[0120] The fault cause investigation module M3 is used to gradually input the current inspection phenomenon based on the multiple rounds of dialogue-based fault diagnosis feedback and conduct fault cause investigation.

[0121] The fault diagnosis result generating module M4 is used to generate a fault diagnosis result after the fault cause is checked. The fault diagnosis result includes: the subsequent handling of the fault and the corrective measures.

[0122] As an optional implementation, the substation equipment fault diagnosis device based on the inspection thought chain includes a maintenance thought chain data construction module, a function call registration module, a large model fine-tuning training module, and an interactive dialogue module. The maintenance thought chain data construction module is connected to the large model fine-tuning training module, the function call registration module is connected to the large model fine-tuning training module, and the large model fine-tuning training module is connected to the interactive dialogue module. The maintenance thought chain data construction module constructs a maintenance thought chain based on the content of the substation equipment fault report using a step-by-step thinking thought chain approach. Each node in the constructed thought chain represents a round of dialogue data, primarily including questions and answers such as the maintenance method, maintenance results, and maintenance conclusions for the current step.

[0123] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the ledger information and abnormal phenomena of the device under test. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for diagnosing faults of substation equipment based on the inspection thinking chain is realized.

[0124] Those skilled in the art will understand that Figure 5 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0125] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the above method embodiments when executing the computer program.

[0126] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the above-mentioned method embodiments when executed by a processor.

[0127] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the above method embodiments are implemented.

[0128] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0129] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0130] The database involved in each embodiment provided by the present invention may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchain. The processor involved in each embodiment provided by the present invention may be, but is not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, etc.

[0131] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0132] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for fault diagnosis of substation equipment based on inspection thinking chain, characterized in that: The power transformation equipment fault diagnosis method based on the inspection thinking chain includes: Obtain the record information and abnormal phenomena of the equipment under test; Inputting the equipment under test's ledger information and the abnormal phenomenon into a fault diagnosis model to obtain multi-round conversational fault diagnosis feedback from the fault diagnosis model; the fault diagnosis model is a model trained based on a large model; the multi-round conversational fault diagnosis feedback includes the fault cause investigation process; Based on the multi-round dialogue-based fault diagnosis feedback, the current inspection phenomenon is gradually input to conduct fault cause investigation; After the cause of the fault is checked, a fault diagnosis result is generated; the fault diagnosis result includes: the subsequent handling of the fault and corrective measures.

2. The method for diagnosing faults of substation equipment based on the inspection thinking chain according to claim 1 is characterized in that: The determination process of the fault diagnosis model is as follows: Based on historical fault repair reports, determine the inspection chain and function registration data for different faults; the inspection chain includes: equipment ledger information, fault phenomenon information, fault inspection process and fault handling; the function registration data is the corresponding external function or service called according to the instruction, which is used to integrate various professional tools involved in the maintenance process; Construct a fine-tuning dataset based on the inspection thinking chain and the function registration data; the fine-tuning dataset includes: fault ledger information, fault abnormality, fault cause investigation process, subsequent processing status and corrective measures; Annotating the fine-tuning dataset to obtain an annotated multi-round conversational fine-tuning dataset; The fault log information and fault anomalies in the labeled multi-round conversational fine-tuning dataset are used as input, and the fault cause investigation process, subsequent processing status and corrective measures corresponding to the fault log information and fault anomalies are used as output. The large model is trained through multi-round conversational fine-tuning to obtain a fault diagnosis model.

3. The method for diagnosing faults of substation equipment based on the inspection thinking chain according to claim 2 is characterized in that: The process of determining the function registration data is as follows: Using a large model, based on user input information, it is determined whether a function needs to be called and the judgment result is obtained; If the judgment result is yes, then determine the function to be called and the corresponding parameters; Generate a structured output according to the function to be called and the corresponding parameters; the structured output includes the parameters required for the function call; According to the structured output, calling a corresponding function; Function registration data is obtained according to the corresponding function.

4. The method for diagnosing faults of substation equipment based on the inspection thinking chain according to claim 2 is characterized in that: The fine-tuning dataset is labeled to obtain a labeled multi-round conversational fine-tuning dataset, specifically including: Construct prompt words; The fine-tuning dataset is manually filled with the prompt words to obtain a labeled multi-round conversational fine-tuning dataset.

5. The method for diagnosing faults of substation equipment based on the inspection thinking chain according to claim 1 is characterized in that: The power transformation equipment fault diagnosis method based on the inspection thinking chain also includes: A fault diagnosis report is generated according to the fault diagnosis result.

6. The method for diagnosing faults of substation equipment based on the inspection thinking chain according to claim 5 is characterized in that: The power transformation equipment fault diagnosis method based on the inspection thinking chain also includes: The fault diagnosis report is output to the user or expert via a user interface or interface.

7. A fault diagnosis device for substation equipment based on inspection thinking chain, characterized in that: The power transformation equipment fault diagnosis device based on the inspection thinking chain includes: Data acquisition module, used to obtain the ledger information and abnormal phenomena of the equipment under test; An input module is configured to input the equipment under test's ledger information and the abnormal phenomenon into a fault diagnosis model to obtain multi-round conversational fault diagnosis feedback from the fault diagnosis model; the fault diagnosis model is a model trained based on a large model; the multi-round conversational fault diagnosis feedback includes the fault cause investigation process; A fault cause troubleshooting module is used to gradually input the current inspection phenomenon based on the multiple rounds of dialogue-based fault diagnosis feedback to troubleshoot the cause of the fault; The fault diagnosis result generating module is used to generate the fault diagnosis result after the fault cause is found. The fault diagnosis result includes the subsequent handling of the fault and the corrective measures.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the substation fault diagnosis method based on the inspection thinking chain described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for diagnosing faults of substation equipment based on an inspection thinking chain according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for diagnosing faults of substation equipment based on an inspection thinking chain according to any one of claims 1 to 6 is implemented.

Images