High-voltage circuit breaker fault diagnosis method, system, medium and equipment
By wavelet transforming the vibration signal of the high-voltage circuit breaker to generate a time-frequency diagram, and combining with a large language model to perform multiple rounds of Q&A interaction, the problem of low fault diagnosis efficiency of high-voltage circuit breakers in the existing technology is solved, and efficient and interpretable fault identification and diagnosis are achieved.
Patent Information
- Application Number
- CN202510620710.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-14
- Publication Date
- 2025-08-12
AI Technical Summary
The existing high-voltage circuit breaker fault diagnosis methods rely on expert experience, are inefficient and subjective, making it difficult to efficiently diagnose various fault types.
By obtaining the vibration signal of the high-voltage circuit breaker, wavelet transformation is performed to generate time-frequency diagrams, and using a large language model to perform multiple rounds of Q&A interaction, combined with the optimization of expert diagnosis results, we can realize automatic identification and diagnosis of fault characteristics.
It improves the objectivity and accuracy of fault diagnosis, reduces manual intervention, enhances the interpretability and efficiency of the diagnostic process, and continuously improves with model optimization.
Smart Images
Figure CN120468634A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a high-voltage circuit breaker fault diagnosis method, system, medium and equipment. Background Art
[0002] High-voltage circuit breakers are among the most critical protection and control devices in power systems, and their reliability is directly linked to the safe operation of the power grid. Statistics show that mechanical failures account for over 70% of all high-voltage circuit breaker failures. Traditional fault diagnosis methods rely primarily on regular inspections and manual judgment, which suffer from low efficiency and high subjectivity.
[0003] In recent years, circuit breaker fault diagnosis methods based on vibration signal analysis have been widely studied. These methods typically collect vibration signals from circuit breaker operation, extract features in the time domain, frequency domain, or time-frequency domain, and then use machine learning algorithms for classification and diagnosis. However, these methods have limitations: feature extraction relies on expert experience, and different faults may require different feature extraction methods. This limitation can result in an incomplete representation of the circuit breaker status, making it difficult to efficiently diagnose high-voltage circuit breaker faults. Summary of the Invention
[0004] The present invention provides a high-voltage circuit breaker fault diagnosis method, system, medium, and device to solve the above-mentioned problem existing in the prior art, namely, how to improve the fault diagnosis efficiency of high-voltage circuit breakers in the prior art. The present invention provides a high-voltage circuit breaker fault diagnosis method, which includes:
[0005] Obtain vibration signals of high-voltage circuit breakers under different fault conditions;
[0006] Perform wavelet transform on the vibration signal to generate the corresponding time-frequency graph;
[0007] Obtain the fault description text corresponding to the time-frequency graph and input it and the time-frequency graph into the large language model for training;
[0008] The time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed is input into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
[0009] Optionally, performing wavelet transform processing on the vibration signal to generate a corresponding time-frequency graph specifically includes:
[0010] By selecting the Morlet wavelet basis function, the vibration signal x(t) is transformed by continuous wavelet transform, and the time-frequency distribution is obtained using the following formula:
[0011]
[0012] Among them, Ψ(t) is the Morlet wavelet basis function, a is the scale factor, and b is the translation parameter.
[0013] Optionally, the different fault states specifically include:
[0014] Normal state, mechanical jamming fault, spring breakage fault, contact wear fault and mechanism loose fault.
[0015] Optionally, the large language model is a LlaMA model.
[0016] Optionally, the fault diagnosis result is compared with the expert diagnosis result to optimize the large language model.
[0017] Optionally, before performing wavelet transform processing on the vibration signal, the vibration signal is subjected to denoising and normalization processing; and the time-frequency graph is subjected to Gaussian noise enhancement, random cropping and edge sharpening processing.
[0018] The present invention provides a high-voltage circuit breaker fault diagnosis system, comprising:
[0019] A signal acquisition module is used to obtain vibration signals of high-voltage circuit breakers under different fault conditions;
[0020] A signal processing module is used to perform wavelet transform on the vibration signal and generate a corresponding time-frequency graph;
[0021] The model training module is used to obtain the fault description text corresponding to the time-frequency graph and input it and the time-frequency graph into the large language model for training;
[0022] The interactive diagnosis module is used to input the time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
[0023] Optionally, the system further includes:
[0024] The feedback optimization module is used to compare the fault diagnosis result with the expert diagnosis result to optimize the large language model.
[0025] The present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the high-voltage circuit breaker fault diagnosis method is implemented.
[0026] 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. When the processor executes the program, the high-voltage circuit breaker fault diagnosis method is implemented.
[0027] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention provides a high-voltage circuit breaker fault diagnosis method, which reduces the reliance on expert feature extraction experience by directly learning multiple types of fault features from time-frequency graphs. It reduces the degree of manual intervention, improves the objectivity and accuracy of the diagnosis process, and can effectively diagnose various types of faults in high-voltage circuit breakers, thereby improving the efficiency of fault diagnosis. At the same time, the present invention performs fault diagnosis through natural language interaction, improves the interpretability of the diagnosis process, makes the cause of the fault more intuitive and easy to understand, and improves the interpretability of fault diagnosis. In addition, by combining the generalization ability of the large language model, with the increase of training samples and the continuous optimization of the model, the accuracy and adaptability of the diagnosis can be continuously improved, thereby improving the efficiency of fault diagnosis, and providing a new idea for intelligent diagnosis of power equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0029] Figure 1 A flowchart of a high-voltage circuit breaker fault diagnosis provided by an embodiment of the present invention;
[0030] Figure 2 A schematic diagram of the process of converting a vibration signal into a time-frequency diagram according to an embodiment of the present invention;
[0031] Figure 3 A schematic diagram of the process of large language model training and question-answering diagnosis provided by an embodiment of the present invention;
[0032] Figure 4 A structural block diagram of a high-voltage circuit breaker fault diagnosis system provided by an embodiment of the present invention;
[0033] Figure 5 A schematic diagram of a computer device for a high-voltage circuit breaker fault diagnosis method provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0034] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.
[0035] The following describes in detail the technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems using specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The following embodiments of the present invention are described in conjunction with the accompanying drawings.
[0036] Figure 1 FIG. 1 is a flow chart of a high-voltage circuit breaker fault diagnosis method provided by an embodiment of the present invention. Figure 1 As shown, this embodiment shows a high-voltage circuit breaker fault diagnosis method, including:
[0037] S1: Obtain vibration signals of high-voltage circuit breaker under different fault conditions.
[0038] For example, vibration signals from high-voltage circuit breakers can be collected under different fault conditions using, for example, accelerometers and data acquisition cards. A variety of typical fault conditions can be set up in the lab or on-site, such as normal state, mechanical jamming, spring breakage, contact wear, and loose mechanism. When using accelerometers to collect vibration signals during circuit breaker opening and closing operations, the sampling frequency should be no less than 50 kHz to ensure signal integrity.
[0039] S2: Perform wavelet transform on the vibration signal to generate the corresponding time-frequency diagram.
[0040] like Figure 2 As shown in the figure, after preprocessing the collected vibration signal, a time-frequency diagram is generated through wavelet transform. Specifically, it includes:
[0041] Signal preprocessing: De-noise the original vibration signal using the wavelet threshold denoising method; then perform normalization to eliminate amplitude differences.
[0042] Wavelet transform: Select a suitable wavelet basis function (such as Morlet wavelet) and perform continuous wavelet transform (CWT) on the vibration signal x(t) to obtain the time-frequency distribution. The specific formula is as follows:
[0043]
[0044] Among them, Ψ(t) is the Morlet wavelet basis function, a is the scale factor, and b is the translation parameter.
[0045] For example, the generated time-frequency graph may be subjected to processing such as adding Gaussian noise enhancement, random cropping, and edge sharpening to improve feature recognizability.
[0046] S3: Obtain the fault description text corresponding to the time-frequency graph, and input it and the time-frequency graph into the large language model for training.
[0047] like Figure 3 As shown in the figure, a training dataset is constructed and a large language model is trained to understand the time-frequency graph features. Specifically, a detailed text description is provided for the time-frequency graph of each fault state. The description should include the main features of the time-frequency graph and its association with the fault state. For example: "The time-frequency graph shows a clear high-frequency component at 0.5s on the time axis, with a significantly increased amplitude, corresponding to the characteristic vibration of a spring break fault." A multimodal large language model (such as LlaMA) is used as input for training, and the time-frequency graph and its text description are input. Through training with a large number of samples, the model learns to associate the visual features of the time-frequency graph with the fault type.
[0048] The structure of the multimodal large language model is divided into three parts:
[0049] (1) Vision Encoder
[0050] Use pre-trained visual models (such as CLIP-ViT, ResNet) to extract global and local features of the time-frequency graph. Input time-frequency graph Generate visual feature vectors through visual encoder
[0051] V=VisionEncoder(I)
[0052] For example, the ViT model divides the image into N patches and extracts features through the Transformer layer: V = [v cls ,v1,...,v N ],
[0053] (2) Text Encoder
[0054] Use the word embedding layer of a large language model (such as LLaMA's Tokenizer) to convert the fault description text T into a word vector sequence {w1,…,w M}.
[0055] (3) Multimodal fusion module
[0056] Align visual and text features through the cross-attention mechanism. For the i-th text word vector wi, its interaction with the visual feature is expressed as:
[0057]
[0058] Where Q, K are the learnable query and key matrices, and dk is the dimensionality scaling factor.
[0059] For example, the training of a large multimodal language model can be divided into two stages:
[0060] (1) Pre-training stage
[0061] Align visual and textual features using contrastive learning loss:
[0062]
[0063] Where s(V,T) is the visual-text similarity score, τ is the temperature coefficient, and B is the batch sample.
[0064] (2) Instruction fine-tuning stage
[0065] The model is trained on the time-frequency graph and fault description pairs to generate diagnostic text. The loss function is the autoregressive language modeling loss:
[0066]
[0067] For example, if you input a time-frequency graph and the prompt "Describe the fault characteristics of this time-frequency graph," the model needs to generate text similar to "High-frequency oscillation (800 Hz) appears at 0.3s on the time axis, with an amplitude 20dB higher than the normal state, which is consistent with the characteristics of a spring break."
[0068] Prompt Engineering: Design and optimize prompt templates to improve the model's ability to describe and analyze time-frequency graph features.
[0069] For example, for fault diagnosis tasks, this application designs the following hierarchical prompt template to guide model reasoning:
[0070] (1) Feature extraction tips
[0071] Please analyze the following characteristics of this time-frequency graph step by step: 1. Frequency distribution within the time axis [Xs, Ys] interval; 2. Regions with abnormal amplitude changes; 3. Differences from normal time-frequency graphs.
[0072] (2) Fault inference prompts
[0073] Based on the following features: Feature 1: {Feature 1 extracted from the model}; Feature 2: {Feature 2 extracted from the model}, please determine the most likely fault type and select from the following options: [A] Mechanical jamming [B] Spring breakage [C] Contact wear [D] Mechanism loosening
[0074] (3) Confirmatory prompts
[0075] "Do you believe the spring is broken based on the following characteristics: - A sudden increase in the amplitude of the high-frequency component (>500Hz)? - A prolonged vibration duration?"
[0076] S4: Input the time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
[0077] For example, the vibration signal of the circuit breaker to be diagnosed is converted into a time-frequency graph, fed into a trained large language model, and diagnostic results are obtained through a question-and-answer method. Specifically, this involves: 1. Initial analysis: Inputting the time-frequency graph allows the model to describe its key features. 2. Feature querying: Guiding the model to focus on key features through specific questions, such as "Is the high-frequency component at time axis X abnormal?" 3. Fault inference: Based on feature analysis, the model infers the most likely fault type. 4. Result verification: Verifying the reliability of the diagnostic results through multiple rounds of question-and-answer sessions.
[0078] For example, the model's diagnostic results can be compared with expert diagnosis to calculate indicators such as accuracy. Based on the evaluation results, the model parameters and question-answering strategy are adjusted to continuously optimize the diagnostic performance. The fault type probability distribution output by the model is P(y|V), and the top-1 probability is taken as the confidence level:
[0079] Confidence = max y∈Y P(y|V),Y={fault A, fault B, ...}
[0080] The model's diagnostic results are as follows: "The time-frequency graph shows sustained high-frequency oscillation (approximately 750 Hz) at 0.2 s, with an amplitude 15 dB higher than normal, and the vibration duration is extended by 30 ms, which is consistent with the typical characteristics of a spring breakage failure (93.7% confidence level)."
[0081] The above is a high-voltage circuit breaker fault diagnosis method provided by one or more embodiments of this specification. Based on the same idea, this specification also provides a corresponding high-voltage circuit breaker fault diagnosis system, such as Figure 4 shown.
[0082] Figure 4 A schematic diagram of a high-voltage circuit breaker fault diagnosis system provided for this specification includes:
[0083] A signal acquisition module is used to obtain vibration signals of high-voltage circuit breakers under different fault conditions;
[0084] A signal processing module is used to perform wavelet transform on the vibration signal and generate a corresponding time-frequency graph;
[0085] The model training module is used to obtain the fault description text corresponding to the time-frequency graph and input it and the time-frequency graph into the large language model for training;
[0086] The interactive diagnosis module is used to input the time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
[0087] Optionally, the system further comprises:
[0088] The feedback optimization module is used to compare the fault diagnosis result with the expert diagnosis result to optimize the large language model.
[0089] The specific limitations of the high-voltage circuit breaker fault diagnosis system can be found in the limitations of the high-voltage circuit breaker fault diagnosis method described above and will not be repeated here. Each module in the above-mentioned high-voltage circuit breaker fault diagnosis system can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.
[0090] The present invention also provides a computer-readable storage medium, which stores a computer program. The computer program can be used to execute the high-voltage circuit breaker fault diagnosis method provided above.
[0091] The present invention also provides Figure 5 The structural diagram of the computer equipment shown in FIG. Figure 5 As shown, at the hardware level, the computer device includes a processor, an internal bus, a network interface, memory, and non-volatile storage, and may also include other hardware required for its operations. The processor reads the corresponding computer program from the non-volatile storage into the internal memory and then runs it to implement the high-voltage circuit breaker fault diagnosis method provided in the above embodiment.
[0092] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods 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, storage, 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 or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0093] The technical features of the above embodiments can be combined arbitrarily. In order 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 the present invention.
Claims
1. A high voltage circuit breaker fault diagnosis method, characterized in that: include: Obtain vibration signals of high-voltage circuit breakers under different fault conditions; Perform wavelet transform on the vibration signal to generate the corresponding time-frequency graph; Obtain the fault description text corresponding to the time-frequency graph and input it and the time-frequency graph into the large language model for training; The time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed is input into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
2. The high-voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The wavelet transform processing of the vibration signal to generate the corresponding time-frequency graph specifically includes: By selecting the Morlet wavelet basis function, the vibration signal x(t) is transformed by continuous wavelet transform, and the time-frequency distribution is obtained using the following formula: Among them, Ψ(t) is the Morlet wavelet basis function, a is the scale factor, and b is the translation parameter.
3. The high-voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The different fault states specifically include: Normal state, mechanical jamming fault, spring breakage fault, contact wear fault and mechanism loose fault.
4. The high-voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The large language model is the LlaMA model.
5. The high-voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: The fault diagnosis result is compared with the expert diagnosis result to optimize the large language model.
6. The high-voltage circuit breaker fault diagnosis method according to claim 1, characterized in that: Before the vibration signal is subjected to wavelet transform processing, the vibration signal is subjected to denoising and normalization processing; and the time-frequency graph is subjected to Gaussian noise enhancement, random cropping and edge sharpening processing.
7. A high voltage circuit breaker fault diagnosis system, characterized in that: include: A signal acquisition module is used to obtain vibration signals of high-voltage circuit breakers under different fault conditions; A signal processing module is used to perform wavelet transform on the vibration signal and generate a corresponding time-frequency graph; The model training module is used to obtain the fault description text corresponding to the time-frequency graph and input it and the time-frequency graph into the large language model for training; The interactive diagnosis module is used to input the time-frequency graph corresponding to the vibration signal of the high-voltage circuit breaker to be diagnosed into the trained large language model. Through multiple rounds of question-and-answer interaction, the time-frequency graph features are determined and described to obtain the fault diagnosis results.
8. The high-voltage circuit breaker fault diagnosis system according to claim 7, characterized in that: Also includes: The feedback optimization module is used to compare the fault diagnosis result with the expert diagnosis result to optimize the large language model.
9. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by a processor, the high-voltage circuit breaker fault diagnosis method according to any one of claims 1 to 6 is implemented.
10. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for diagnosing a high-voltage circuit breaker fault according to any one of claims 1 to 6 is implemented.