Health assessment method based on environmental awareness multi-modal dynamic fusion measurement assets
By acquiring multimodal environmental data for feature fusion and adaptive training, the problem of fault diagnosis accuracy of power grid metering assets under complex operating conditions has been solved, achieving higher fault identification rate and health assessment accuracy.
Patent Information
- Application Number
- CN202511138553.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-14
- Publication Date
- 2025-10-31
AI Technical Summary
Existing multimodal visual learning models suffer from reduced generalization performance and insufficient dynamic perception under complex operating conditions such as lighting interference and low illumination in foggy weather at power grid metering asset sites, resulting in low fault diagnosis accuracy.
By acquiring environmental parameter data, including visible light, infrared thermal phase, and ambient humidity data, feature fusion and adaptive training are performed to build a failure probability model, assess the health status of the measured assets, adjust weights in real time to filter interference, and improve system robustness.
It significantly improves the accuracy of fault diagnosis under complex interference, increases the fault identification rate and the accuracy of health status assessment, and reduces the false alarm rate and the model's dependence on labeled data.
Smart Images
Figure CN120875447A_ABST
Abstract
Description
Technical Field
[0001] This invention generally relates to the field of intelligent operation and maintenance of power equipment, and more specifically, to a health assessment method for metering assets based on environmental perception and multimodal dynamic fusion. Background Technology
[0002] In recent years, with the intelligent transformation of power grids, real-time status monitoring and accurate fault diagnosis of metering assets have become core aspects of improving quality monitoring.
[0003] Significant breakthroughs have been achieved in the theoretical innovation of multimodal visual learning, with cross-modal alignment mechanisms emerging as a key research direction, yielding numerous important results. For example, the multimodal cueing collaborative network model optimizes semantic-image alignment in power equipment inspection through parallel feature interaction between the language encoder and visual encoder, significantly improving multi-task detection accuracy; the attribute-class contrast loss algorithm, by mining the discriminative attributes of visual objects, achieves an average recognition accuracy of 76.84% on fine-grained datasets such as Stanford Dog-120, a 10.89% improvement over traditional models. Furthermore, existing self-supervised learning systems summarize three major objective functions: instance discrimination, mask prediction, and clustering, significantly reducing the model's dependence on labeled data and laying a theoretical foundation for the utilization of unlabeled multi-source data in industrial scenarios.
[0004] However, most existing research models rely on well-structured data in laboratory environments, while power grid metering assets often face complex operating conditions such as light interference and low illumination in foggy weather, which reduces the generalization performance of the models. Moreover, existing research lacks dynamic perception. Summary of the Invention
[0005] According to the present invention, a health assessment scheme based on environmentally perceptive multimodal dynamic fusion of measurement assets is provided. This scheme effectively solves the problem of lag in traditional static fusion response through real-time weight adjustment and interference filtering mechanisms, significantly improving system robustness.
[0006] In a first aspect of the invention, a health assessment method for metering assets based on environmentally perceptual multimodal dynamic fusion is provided. The method includes: Acquire environmental parameter data, which includes visible light input data, infrared thermal phase input data, ambient humidity data, and status index data; The visible light input data is normalized to obtain illumination data; the infrared thermal phase input data is preprocessed to obtain infrared thermal phase data. Based on the illumination data, infrared thermal phase data, and ambient humidity data, illumination feature vector, infrared thermal phase feature vector, and ambient feature vector are calculated. The illumination feature vector, infrared thermal feature vector, and environmental feature vector are fused to obtain a multimodal feature vector; A failure probability model is constructed, and the failure probability model is adaptively trained using the state index data. The multimodal feature vector is then input into the trained failure probability model to obtain a confidence score, and the health status of the measured asset is assessed using the confidence score.
[0007] Further, the visible light input data is preprocessed to obtain illumination data, including:
[0008]
[0009] in, This is lighting data; Input data for visible light; This is the first illumination compensation coefficient; Light intensity; This represents the basic light intensity. This is the second illumination compensation coefficient; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
[0010] Further, the preprocessing of the infrared thermal phase input data to obtain infrared thermal phase data includes:
[0011]
[0012] in, This is lighting data; Input data for visible light; This is the first illumination compensation coefficient; Light intensity; This represents the basic light intensity. This is the second illumination compensation coefficient; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
[0013] Further, based on the illumination data, infrared thermal phase data, and ambient humidity data, illumination feature vectors, infrared thermal phase feature vectors, and environmental feature vectors are calculated, including:
[0014]
[0015]
[0016]
[0017]
[0018] in, This is the first infrared thermal image feature vector; This is the first learnable weight matrix; The second infrared thermal image feature vector is given by; AvgPool is the average pooling operation. Infrared thermal phase data; Sigmoid is a sigmoid function; This is the second learnable weight matrix; For thermal anomaly region mask; First illumination feature vector; Light intensity; The x-axis represents the light intensity. The y-axis represents the light intensity. For integration operations; Second illumination feature vector; Statistics for channel histograms; For color space conversion; This is lighting data; For environmental feature vectors; The feature fusion weight matrix; This is a scalar value for ambient humidity. The rate of change of humidity; Enter the humidity value.
[0019] Further, the feature fusion of the illumination feature vector, infrared thermal feature vector, and environmental feature vector to obtain a multimodal feature vector includes:
[0020] in, It is a multimodal feature vector; This is the first infrared thermal image feature vector; The second infrared thermal image feature vector is: This is the first illumination feature vector; This is the second illumination feature vector; This is the environmental feature vector.
[0021] Furthermore, the adaptive training of the fault probability model using the state index data includes: The state index data is normalized to obtain normalized state index data. The health index is calculated based on the normalized state index data. The health index is input into the failure probability model, and the failure probability model is adaptively trained.
[0022] Further, the normalization process for the state index data to obtain normalized state index data includes:
[0023]
[0024]
[0025]
[0026] in, Entropy factor; The coefficients are the first AHP-entropy weighting coefficients; The coefficients are the second AHP-entropy weighting coefficients; AHP-entropy weighting factor; The coefficients are the third AHP-entropy weighting method coefficients; AHP factor; For the first One input index vector; Corrosion rate; This is the resistance value; This is the temperature difference value; This is a normalized state indicator.
[0027] Furthermore, the calculation of the health index based on the normalized state index data includes:
[0028] in, As a health index; The total number of indicators; For indexing indicators.
[0029] Further, the step of inputting the multimodal feature vector into the trained fault probability model to obtain the confidence score includes:
[0030] in, Confidence level; Fault type; For the first The weights corresponding to different types of faults; For the first The linear constants corresponding to each type of fault.
[0031] In a second aspect of the invention, an electronic device is provided. The electronic device includes at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of the first aspect of the invention.
[0032] Compared with the prior art, the present invention has the following beneficial technical effects: the present invention can significantly improve the accuracy of fault diagnosis under complex interference.
[0033] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0034] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A flowchart of a health assessment method for environmentally perceptive multimodal dynamic fusion measurement assets based on an embodiment of the present invention is shown. Figure 2 A block diagram of fault model training according to an embodiment of the present invention is shown; Figure 3 A block diagram of an exemplary electronic device capable of implementing embodiments of the present invention is shown.
[0035] Among them, 300 is an electronic device, 301 is a computing unit, 302 is a ROM, 303 is a RAM, 304 is a bus, 305 is an I / O interface, 306 is an input unit, 307 is an output unit, 308 is a storage unit, and 309 is a communication unit. Detailed Implementation
[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0037] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0038] In this invention, environmental parameter data is acquired, and feature vectors are calculated. These feature vectors are then fused to obtain a multimodal feature vector. This multimodal feature vector is input into a trained fault probability model to obtain a confidence score. The health status of the measured asset is then assessed using this confidence score. This method enables accurate quantitative assessment of the health status of the measured asset.
[0039] Figure 1 A flowchart of a health assessment method for metered assets based on environmental perception and multimodal dynamic fusion, according to an embodiment of the present invention, is shown.
[0040] The method includes: S101. Obtain environmental parameter data, which includes visible light input data, infrared thermal phase input data, ambient humidity data, and status index data.
[0041] Specifically, environmental parameter data are acquired through light sensors, infrared sensors, and humidity sensors.
[0042] S102. Normalize the visible light input data to obtain illumination data; preprocess the infrared thermal phase input data to obtain infrared thermal phase data.
[0043] In this embodiment, the visible light input data is preprocessed to normalize it to obtain illumination data, including:
[0044]
[0045] in, This is lighting data; Input data for visible light; This is the first illumination compensation coefficient; Light intensity; This represents the basic light intensity. This is the second illumination compensation coefficient; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
[0046] By preprocessing the visible light input data, it is possible to identify and process interference data caused by random fluctuations or erroneous measurements, making the data more reflective of the true pattern.
[0047] In this embodiment, the Stefan-Boltzmann law is used to suppress thermal noise in the environment, thereby preprocessing the infrared thermal phase input data to obtain infrared thermal phase data, including:
[0048]
[0049] in, Infrared thermal phase data; Input data for infrared thermal phase; The coefficient of the first Stefan-Boltzmann law; The coefficient of the second Stefan-Boltzmann law; The ambient temperature; The basic ambient temperature; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
[0050] By preprocessing the infrared thermal phase input data, duplicate records can be identified, deleted, or merged, thus avoiding data redundancy that could lead to biased analysis results.
[0051] S103. Based on the illumination data, infrared thermal phase data, and ambient humidity data, calculate the illumination feature vector, infrared thermal phase feature vector, and ambient feature vector, including: ; ; ; ; ;
[0052] in, This is the first infrared thermal image feature vector; This is the first learnable weight matrix; The second infrared thermal image feature vector is given by; AvgPool is the average pooling operation. Infrared thermal phase data; Sigmoid is a sigmoid function; This is the second learnable weight matrix; For thermal anomaly region mask; First illumination feature vector; Light intensity; The x-axis represents the light intensity. The y-axis represents the light intensity. For integration operations; Second illumination feature vector; Statistics for channel histograms; For color space conversion; This is lighting data; For environmental feature vectors; The feature fusion weight matrix; This is a scalar value for ambient humidity. The rate of change of humidity; Enter the humidity value.
[0053] By calculating the above five feature vectors, we can identify and retain the features that contribute most to the prediction of the target variable, eliminate redundant or irrelevant features, simplify the model, reduce the risk of overfitting, and improve interpretability.
[0054] S104. The illumination feature vector, infrared thermal phase feature vector, and environmental feature vector are fused to obtain a multimodal feature vector, including:
[0055] in, It is a multimodal feature vector; This is the first infrared thermal image feature vector; The second infrared thermal image feature vector is: This is the first illumination feature vector; This is the second illumination feature vector; This is the environmental feature vector.
[0056] Different feature vectors can describe the same object from different perspectives. By fusing all feature vectors, we can make comprehensive use of these complementary information and form a more comprehensive representation.
[0057] S105. Construct a failure probability model, adaptively train the failure probability model using the state index data, input the multimodal feature vector into the trained failure probability model to obtain the confidence level, and evaluate the health status of the measured asset using the confidence level.
[0058] In this embodiment, as Figure 2 As shown, the adaptive training of the fault probability model using the state index data includes: S201. Normalize the state index data to obtain normalized state index data.
[0059] Specifically, for multivariate indicators, the combined weights are determined using a combination of AHP and entropy weighting methods. The AHP weights are determined by constructing a judgment matrix through expert scoring, including:
[0060]
[0061] in, Entropy factor; The coefficients are the first AHP-entropy weighting coefficients; The coefficients are the second AHP-entropy weighting coefficients; AHP-entropy weighting factor; The coefficients are the third AHP-entropy weighting method coefficients; It is an AHP factor.
[0062] Specifically, for the input indicators of the fuzzy state evaluation, the state level is calculated through the fuzzy membership function, including:
[0063]
[0064] in, For the first One input index vector; Corrosion rate; This is the resistance value; This is the temperature difference value; This is a normalized state indicator.
[0065] Normalizing status indicator data refers to merging and transforming data obtained from different sources, formats, and standards to make them compatible under the same analytical framework, thereby ensuring that all data follow the same rules and facilitating subsequent processing and analysis.
[0066] Specifically, the status indicator data are shown in Table 1.
[0067] Table 1
[0068] in, Corrosion rate; Minimum corrosion rate; The maximum corrosion rate; Insulation resistance; Minimum insulation resistance; Maximum insulation resistance; Assess the protection level rating; For coefficients; It is waterproof and dustproof. For coefficients; This is an overheating risk index; These are constant coefficients.
[0069] S202. Based on the normalized state index data, a health index is calculated, including:
[0070] The health index is calculated by normalizing state indicator data, which can improve the credibility of conclusions drawn from normalized and consistent data.
[0071] Specifically, the fault-state mapping relationship of the metering box is shown in Table 2.
[0072] Table 2
[0073] in, As a health index.
[0074] S203. Input the health index into the failure probability model and perform adaptive training on the failure probability model.
[0075] In this embodiment, the multimodal feature vector is input into the trained fault probability model to obtain the confidence level, including:
[0076] in, Confidence level; Fault type =1,2,3 correspond to enclosure, electrical, and environmental faults, respectively; For the first The weights corresponding to different types of faults; For the first The linear constants corresponding to each type of fault.
[0077] Specifically, assessing the health status of the measured assets through the confidence level includes:
[0078] in, In good health; For health; For early warning; It is unhealthy.
[0079] The fault diagnosis performance of the present invention compared with that of the prior art is shown in Table 4.
[0080] Table 4
[0081] As shown in Table 4, in the detection of enclosure corrosion faults, the recall rate of the algorithm proposed in this invention is increased to 95.2% (12.1% higher than the DCMF algorithm) due to the enhanced infrared features of the dynamic gating under low illumination; the false alarm rate of electrical overheating faults is reduced to 3.1% (67% lower than the baseline), thanks to the joint frequency-spatial filtering.
[0082] The state assessment performance of the present invention compared with that of the prior art is shown in Table 5.
[0083] Table 5
[0084] Table 5 shows the assumed corrosion rate The metering box, its true health index: (Warning status). This model predicts: (Error 3.2%), compared to model predictions: (Error 14.3%) The reason for the large error in the comparison model is that the humidity correlation of seal failure was not considered.
[0085] The anti-interference performance of DCMF-Net and the method proposed in this invention is compared in Table 6.
[0086] Table 6
[0087] The effectiveness of the module was verified through ablation experiments, as shown in Table 7.
[0088] Table 7
[0089] The parameter sensitivity analysis in Table 7 shows that the gating network dimension d: when At that time, Ac reached its peak at 93.4% ( , Loss weight temperature coefficient when, MAE is the lowest at ( , ).
[0090] Based on Tables 5-7, the following experimental conclusions can be drawn: 1. Diagnostic accuracy: The fault identification rate reaches 92.4%, which is 8.2% higher than the best comparison model, and has significant advantages, especially in low light and low temperature scenarios.
[0091] 2. Status Assessment: Health Index Prediction The state classification accuracy is 91.7%, achieving accurate quantification of fault decay for the first time.
[0092] 3. Robustness: Under strong interference environment (10^5 lux light intensity / -10°C low temperature), the volatility of key indicators is <5%.
[0093] 4. Engineering Applicability: Edge deployment latency <100ms, providing a highly reliable solution for smart grid operation and maintenance.
[0094] Compared with the prior art, the present invention has the following beneficial technical effects: the present invention can significantly improve the accuracy of fault diagnosis under complex interference.
[0095] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present invention.
[0096] The above is an introduction to the method embodiments. The following describes the solution of the present invention further through an apparatus embodiment that has the same inventive concept as the method in the foregoing embodiments.
[0097] According to embodiments of the present invention, an electronic device is also provided.
[0098] Figure 3 A schematic block diagram of an electronic device 300 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.
[0099] Electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 302 or a computer program loaded from storage unit 308 into random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of electronic device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via bus 304. Input / output (I / O) interface 305 is also connected to bus 304.
[0100] Multiple components in electronic device 300 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of displays, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows electronic device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0101] The computing unit 301 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as methods S101-S105. For example, in some embodiments, methods S101-S105 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of methods S101-S105 described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to execute methods S101 to S105 by any other suitable means (e.g., by means of firmware).
[0102] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0103] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0104] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.
[0105] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A health assessment method for measurement assets based on environmental perception and multimodal dynamic fusion, characterized in that, include: Acquire environmental parameter data, which includes visible light input data, infrared thermal phase input data, ambient humidity data, and status index data; The visible light input data is normalized to obtain illumination data; The infrared thermal phase input data is preprocessed to obtain infrared thermal phase data; Based on the illumination data, infrared thermal phase data, and ambient humidity data, illumination feature vector, infrared thermal phase feature vector, and ambient feature vector are calculated. The illumination feature vector, infrared thermal feature vector, and environmental feature vector are fused to obtain a multimodal feature vector; A failure probability model is constructed, and the failure probability model is adaptively trained using the state index data. The multimodal feature vector is then input into the trained failure probability model to obtain a confidence score, and the health status of the measured asset is assessed using the confidence score.
2. The method according to claim 1, characterized in that, The visible light input data is normalized to obtain illumination data, including: ； ； in, This is lighting data; Input data for visible light; This is the first illumination compensation coefficient; Light intensity; This represents the basic light intensity. This is the second illumination compensation coefficient; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
3. The method according to claim 1, characterized in that, The preprocessing of the infrared thermal phase input data to obtain infrared thermal phase data includes: ； ； in, Infrared thermal phase data; Input data for infrared thermal phase; The coefficient of the first Stefan-Boltzmann law; The coefficient of the second Stefan-Boltzmann law; The ambient temperature; The basic ambient temperature; It is the set of real numbers; The number of rows in the image or spatial grid; The number of columns in the image or spatial grid; This represents the number of channels.
4. The method according to claim 1, characterized in that, Based on the illumination data, infrared thermal phase data, and ambient humidity data, illumination feature vectors, infrared thermal phase feature vectors, and environmental feature vectors are calculated, including: ； ； ； ； ； in, This is the first infrared thermal image feature vector; This is the first learnable weight matrix; The second infrared thermal image feature vector is given by; AvgPool is the average pooling operation. Infrared thermal phase data; Sigmoid is a sigmoid function; This is the second learnable weight matrix; For thermal anomaly region mask; First illumination feature vector; Light intensity; The x-axis represents the light intensity. The y-axis represents the light intensity. For integration operations; Second illumination feature vector; Statistics for channel histograms; For color space conversion; This is lighting data; For environmental feature vectors; The feature fusion weight matrix; This is a scalar value for ambient humidity. The rate of change of humidity; Enter the humidity value.
5. The method according to claim 1, characterized in that, The process of fusing the illumination feature vector, infrared thermal feature vector, and environmental feature vector to obtain a multimodal feature vector includes: ； in, It is a multimodal feature vector; This is the first infrared thermal image feature vector; The second infrared thermal image feature vector is: This is the first illumination feature vector; This is the second illumination feature vector; This is the environmental feature vector.
6. The method according to claim 1, characterized in that, The adaptive training of the fault probability model using the state index data includes: The state index data is normalized to obtain normalized state index data. The health index is calculated based on the normalized state index data. The health index is input into the failure probability model, and the failure probability model is adaptively trained.
7. The method according to claim 6, characterized in that, The normalization process for the state index data to obtain normalized state index data includes: ； ； ； ； in, Entropy factor; The coefficients are the first AHP-entropy weighting coefficients; The coefficients are the second AHP-entropy weighting coefficients; AHP-entropy weighting factor; The coefficients are the third AHP-entropy weighting method coefficients; AHP factor; For the first One input index vector; Corrosion rate; This is the resistance value; This is the temperature difference value; This is a normalized state indicator.
8. The method according to claim 7, characterized in that, The health index is calculated based on the normalized state index data, including: ； in, As a health index; The total number of indicators; For indexing indicators.
9. The method according to claim 5, characterized in that, The step of inputting the multimodal feature vector into the trained fault probability model to obtain the confidence level includes: ； in, Confidence level; Fault type; For the first The weights corresponding to different types of faults; For the first The linear constants corresponding to each type of fault.
10. An electronic device, comprising at least one processor; and a memory communicatively connected to said at least one processor; characterized in that, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-9.