Multi-spectral fusion high-voltage equipment hidden defect detection method and multi-spectral fusion high-voltage equipment hidden defect detection system

Abstract

The invention discloses a multi-spectral fusion high-voltage equipment hidden defect detection method and system, and relates to the technical field of power systems. The method comprises the following steps: collecting multispectral data, and carrying out time alignment and space mapping processing to form a unified spectral feature set; cross-spectrum conflict relation analysis is carried out on the abnormal features, conflict types are identified, and conflict credibility indexes are calculated; constructing a defect energy closed-loop verification model, and calculating an energy closed-loop integrity index; setting a risk budget limit for the target high-voltage equipment, and calculating a risk consumption value to obtain a risk budget balance; determining a current defect stage, switching a current decision dominant spectrum, and generating a dominant right identifier; and when the conflict credibility index, the energy closed-loop integrity index and the dominant right identifier meet corresponding conditions, judging that the target high-voltage equipment has a structural hidden defect, and outputting a corresponding defect type and a risk level. According to the method, the accuracy, the reliability and the mechanism interpretation capability of hidden defect detection are improved.

Description

Technical Field

[0001] This application relates to the field of power system technology, and in particular to a method and system for detecting latent defects in high-voltage equipment using multispectral fusion. Background Technology

[0002] With the continuous expansion of power systems, high-voltage equipment is increasingly widely used in transmission, substation, and distribution systems. During long-term operation, high-voltage equipment is susceptible to various factors such as electrical stress, thermal stress, mechanical stress, and environmental factors, gradually leading to problems such as insulation aging, partial discharge, poor contact, and material deterioration. These problems typically manifest as latent defects in their early stages, with weak external characteristics and slow development, making them difficult to detect in a timely manner through traditional inspections or single detection methods. However, as operating time increases, latent defects may gradually evolve into overt faults, potentially causing equipment damage or even power grid accidents. Therefore, employing multiple spectral detection methods for joint monitoring of high-voltage equipment is crucial for improving the accuracy and reliability of latent defect detection.

[0003] In related technologies, most spectral detection methods only perform simple feature extraction and superposition fusion of different spectral data, lacking in-depth analysis of the intrinsic correlation between various spectral anomalies. This makes it difficult to identify potential logical conflicts or synergistic relationships between different spectral anomalies, leading to misjudgments or missed detections under complex environmental interference. Furthermore, existing methods typically judge anomalies only from the perspective of single features or statistical characteristics, lacking verification mechanisms based on defect energy evolution mechanisms. They cannot comprehensively verify the rationality of anomalies in the energy release, propagation, and deposition processes, resulting in detection results lacking mechanistic support. Therefore, they are significantly insufficient in the early identification of latent defects and long-term risk assessment, and improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for detecting latent defects in high-voltage equipment involving multispectral fusion, so as to solve the problems mentioned in the background art.

[0005] Firstly, this application provides a multispectral fusion method for detecting latent defects in high-voltage equipment, employing the following technical solution: Multispectral data of the target high-voltage equipment are collected within the same detection time window. The multispectral data is then processed by time alignment and spatial mapping to form a unified set of spectral features. Construct a spectral anomaly conflict matrix, perform cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identify conflict types and calculate conflict confidence index; Based on the aforementioned conflict types, a defect energy closed-loop verification model is constructed to verify the closed-loop integrity of the energy release, energy transfer, and energy deposition processes for the corresponding abnormal features, and to calculate the energy closed-loop integrity index. Set a risk budget limit for the target high-voltage equipment, and calculate the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index to obtain the risk budget balance. Based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, the current defect stage is determined, and the current decision-making dominant spectrum is switched according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier; When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the phase consistency condition, it is determined that the target high-voltage equipment has a structural latent defect, and the corresponding defect type and risk level are output.

[0006] Preferably, the step of acquiring multispectral data of the target high-voltage equipment within the same detection time window, and performing time alignment and spatial mapping processing on the multispectral data to form a unified spectral feature set specifically includes: Multispectral data of the target high-voltage equipment are collected within the same detection time window. The multispectral data includes infrared radiation data, ultraviolet radiation data, and visible light data. The timestamp information corresponding to each spectral data is collected respectively. The time offset between each spectral data is calculated based on the timestamp information, and time compensation processing is performed on each spectral data according to the time offset to make each spectral data correspond to the same unified time axis. The image corresponding to the visible light data is selected as the reference coordinate system. Spatial feature matching is performed on the infrared radiation data and ultraviolet radiation data, and the spatial transformation matrix is ​​calculated to map each spectral data to a unified spatial coordinate system. Based on the unified time axis and unified spatial coordinate system, the intensity features and time variation features corresponding to each spectral data are integrated to construct a unified spectral feature set.

[0007] Preferably, the steps of constructing a spectral anomaly conflict matrix, performing cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identifying conflict types, and calculating conflict confidence indices are as follows: The anomalous features corresponding to each spectrum are extracted from the unified spectral feature set. The anomalous features include intensity anomalous features, temporal abrupt change features, and spatial anomalous region features. The anomalous features are then standardized to form an anomalous feature vector set. The abnormal feature vector set is combined in pairs, and the abnormal intensity difference value, response time difference and spatial overlap are calculated respectively to construct the spectral abnormality conflict matrix. Each matrix element in the spectral anomaly conflict matrix is ​​matched with a preset conflict type rule base to identify the corresponding conflict type. The conflict types include intensity-reverse conflict type, time-lag anomaly type, spatial separation type, and cooperative consistency type. Determine whether the conflict type has a corresponding preset interpretation path. If an interpretation path exists, mark it as a structural conflict. Based on the significance of the abnormal intensity and spatial overlap of the structural conflicts, the conflict weights are calculated, and all the structural conflicts are weighted to obtain the conflict credibility index.

[0008] Preferably, the step of calculating conflict weights based on the significance of the abnormal intensity and spatial overlap of the structural conflicts, and then performing a weighted calculation on all the structural conflicts to obtain the conflict credibility index, specifically includes: The conflict units marked as structural conflicts are extracted from the spectral anomaly conflict matrix to form a structural conflict set; For each structural conflict in the set of structural conflicts, the abnormal intensity value of the corresponding spectrum is obtained, the degree of deviation from the standard value is calculated, and the degree of deviation is standardized to obtain the significance of the abnormal intensity. For each structural conflict in the set of structural conflicts, obtain the set of spatial coordinates of the corresponding spectral anomaly region, calculate the cross-union ratio between the anomaly regions, and obtain the spatial overlap. Based on the significance of the anomaly intensity and the degree of spatial overlap, the conflict weights are calculated according to a preset ratio, and the conflict weights of all conflicts in the structural conflict set are summed to obtain the conflict credibility index.

[0009] Preferably, based on the aforementioned conflict type, the steps of constructing a defect energy closed-loop verification model, verifying the closed-loop integrity of the energy release, energy transfer, and energy deposition processes for the corresponding abnormal characteristics, and calculating the energy closed-loop integrity index are as follows: According to the conflict type, the corresponding energy evolution template is invoked, abnormal features are extracted from the unified spectral feature set, and energy release feature parameters, energy transfer feature parameters and energy deposition feature parameters are calculated respectively. Based on the release time in the energy release characteristic parameters, the transfer response time in the energy transfer characteristic parameters, and the precipitation formation time in the energy precipitation characteristic parameters, the time sequence is verified, and the time continuity parameter of energy transfer is calculated. Based on the release location in the energy release characteristic parameters, the transfer path information in the energy transfer characteristic parameters, and the deposition location in the energy deposition characteristic parameters, a spatial relationship analysis is performed to calculate the spatial consistency parameter of energy propagation. Based on the release intensity in the energy release characteristic parameters, the transfer intensity attenuation ratio in the energy transfer characteristic parameters, and the cumulative value of the precipitation intensity in the energy precipitation characteristic parameters, the attenuation rationality parameter of energy attenuation is calculated. The time continuity parameter, spatial consistency parameter, and attenuation rationality parameter are normalized and weighted according to preset weights to obtain the energy closed-loop integrity index.

[0010] Preferably, the step of setting a risk budget for the target high-voltage equipment and calculating the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index to obtain the risk budget balance is as follows: Obtain the equipment operation information of the target high-voltage equipment, and calculate the risk budget amount of the target high-voltage equipment in the current operating cycle according to the preset risk budget model; Extract the conflict confidence index and energy closed-loop integrity index within the current detection time window, and establish corresponding abnormal event records; The conflict credibility index is normalized and combined with a preset conflict type weighting coefficient to calculate the conflict risk consumption value. The energy closed-loop integrity index is normalized and combined with a preset energy evolution weighting coefficient to calculate the energy closed-loop risk consumption value. The conflict risk consumption value and the energy closed-loop risk consumption value are weighted and calculated according to the preset risk fusion weight to obtain the comprehensive risk consumption value. The comprehensive risk consumption value is added to the cumulative risk consumption value of the target high-voltage equipment for updating, and the risk budget balance of the target high-voltage equipment is calculated based on the risk budget amount and the cumulative risk consumption value.

[0011] Preferably, the step of determining the current defect stage based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, and switching the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominance identifier, specifically includes: Extract the conflict confidence index, energy closed-loop integrity index, and risk budget balance for the current detection cycle, and perform a weighted calculation on the risk budget consumption corresponding to the conflict confidence index, energy closed-loop integrity index, and risk budget balance to obtain the defect comprehensive status index. The current defect stage of the target high-voltage equipment is determined by comparing the comprehensive defect status index with a preset defect stage judgment threshold range. According to the preset stage-dominant spectrum mapping rules, the dominant spectrum corresponding to the current defect stage is searched in the stage-dominant spectrum mapping rule library; It is determined whether the dominant spectrum is consistent with the dominant spectrum of the previous detection cycle. If the dominant spectrum is inconsistent with the dominant spectrum of the previous detection cycle, it is determined whether to perform a dominant spectrum switch according to the preset stage stability judgment condition, and a dominant power identifier is generated according to the determined dominant spectrum.

[0012] Preferably, the step of determining that the target high-voltage equipment has a structural latent defect and outputting the corresponding defect type and risk level when the conflict credibility index meets the first threshold condition, the energy closed-loop integrity index meets the second threshold condition, and the dominance identifier meets the stage consistency condition is as follows: Extract the conflict confidence index, energy closed-loop integrity index and dominance identifier of the current detection cycle, and extract the current dominant spectral type and corresponding defect stage information from the dominance identifier; The conflict credibility index is compared with a preset credibility threshold. When the conflict credibility index is greater than or equal to the preset credibility threshold, the abnormal conflict structure is determined to meet the credibility condition. The energy closed-loop integrity index is compared with a preset integrity threshold. When the energy closed-loop integrity index is greater than or equal to the preset integrity threshold, it is determined that the energy closed-loop integrity condition is abnormally met. Based on the defect stage information in the dominant identifier, the consistency of the stage information of the current detection cycle and the historical detection cycle is checked. When the stage is continuously and stably reached a preset number of cycles, it is determined that the stage consistency condition is met. When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, it is determined that the target high-voltage equipment has a structural hidden defect. The preset defect feature library is invoked according to the dominant spectral type, and the current spectral anomaly features are matched and identified to determine the corresponding defect type. Based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, a comprehensive assessment is conducted to determine the risk level of the target high-voltage equipment. Combining the defect type and risk level, the defect detection results are output.

[0013] Secondly, this application provides a multispectral fusion-based high-voltage equipment latent defect detection system, which adopts the following technical solution: A multispectral fusion system for detecting latent defects in high-voltage equipment, comprising: The spectral feature fusion module collects multispectral data of the target high-voltage equipment within the same detection time window, performs time alignment and spatial mapping processing on the multispectral data, and forms a unified set of spectral features. The spectral conflict analysis module constructs a spectral anomaly conflict matrix, performs cross-spectral conflict relationship analysis on the anomalous features in the unified spectral feature set, identifies conflict types, and calculates the conflict confidence index. The energy closed-loop verification module, based on the conflict type, constructs a defective energy closed-loop verification model, verifies the closed-loop integrity of the energy release, energy transfer and energy deposition processes of the corresponding abnormal features, and calculates the energy closed-loop integrity index. The risk budget assessment module sets a risk budget limit for the target high-voltage equipment, calculates the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index, and obtains the risk budget balance. The stage determination and spectrum switching module determines the current defect stage based on the conflict credibility index, energy closed-loop integrity index and risk budget balance, and switches the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier. The defect result output module determines that the target high-voltage equipment has a structural latent defect when the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, and outputs the corresponding defect type and risk level.

[0014] In summary, this application includes at least one of the following beneficial technical effects: 1. By acquiring multispectral data of the target high-voltage equipment within the same detection time window and performing time alignment and spatial mapping on each spectral data, a unified spectral feature set is formed. This allows for collaborative analysis of different spectral information under a unified time axis and spatial coordinate system, eliminating errors caused by differences in acquisition time or spatial perspective between multispectral data and improving the accuracy of multi-source data fusion. Secondly, by constructing a spectral anomaly conflict matrix, cross-spectral conflict relationship analysis is performed on the anomaly features in the unified spectral feature set. Conflict types are identified and conflict credibility indices are calculated, enabling the system to identify logical contradictions between different spectral anomalies, thereby distinguishing between genuine defect anomalies and environmental interference anomalies and improving the credibility of anomaly judgment. Furthermore, a defect energy closed-loop verification model is constructed based on conflict types. The closed-loop integrity of the evolution process of anomaly features in the three stages of energy release, energy transfer, and energy deposition is verified, and the energy closed-loop integrity index is calculated. From the perspective of defect energy evolution mechanism, the physical consistency of anomalies is verified, so that anomaly identification no longer relies solely on a single spectral feature but combines the energy propagation process for comprehensive judgment, thus significantly improving the mechanistic reliability of latent defect identification. Based on this, by setting a risk budget for the target high-voltage equipment and calculating the corresponding risk consumption value according to the conflict credibility index and energy closed-loop integrity index, a risk budget balance is obtained. This allows for dynamic quantification and cumulative management of equipment risk throughout the operating cycle, thereby achieving continuous assessment of potential equipment risks. Furthermore, the current defect stage is determined based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance. The current decision-making dominant spectrum is switched according to a preset stage-dominant spectral mapping rule, ensuring that the most sensitive spectral information dominates the decision-making process at different defect stages. Simultaneously, a stage stability determination mechanism avoids frequent switching, improving the stability of detection decisions. Finally, when the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominant identifier meets the stage consistency condition, a structural latent defect is determined in the target high-voltage equipment. The corresponding defect type and risk level are output, thus achieving accurate identification and risk classification early warning of latent defects in high-voltage equipment, significantly improving the accuracy and stability of latent defect detection.

[0015] 2. By constructing a closed-loop verification model for defect energy based on conflict types, and verifying the closed-loop integrity of abnormal features in the three stages of energy release, energy transfer, and energy deposition, the consistency of multispectral anomalies can be verified from the perspective of the physical evolution mechanism of equipment defects. Specifically, firstly, based on the conflict type, the corresponding energy evolution template is invoked to extract abnormal features from a unified spectral feature set and calculate energy release characteristic parameters, energy transfer characteristic parameters, and energy deposition characteristic parameters, thereby establishing the correspondence between abnormal phenomena and the defect energy evolution process; further, by verifying the time sequence of release time, transfer response time, and deposition formation time, the temporal continuity parameters of energy transfer are obtained, enabling abnormal changes to form a reasonable evolution chain in the time dimension; simultaneously, by analyzing the spatial relationship between release location, transfer path information, and deposition location, the spatial consistency parameters of energy propagation are obtained, verifying the physical correlation between anomalies from the perspective of spatial propagation; in addition, by comprehensively analyzing the release intensity, transfer intensity attenuation ratio, and deposition intensity cumulative value, the attenuation rationality parameters of energy attenuation are calculated to determine whether the changes in energy during the propagation process conform to the defect energy attenuation law. Finally, the time continuity parameter, spatial consistency parameter, and attenuation rationality parameter are normalized and weighted to obtain the energy closed-loop integrity index. This effectively filters out false anomalies caused solely by environmental interference or measurement errors, significantly improving the reliability of the mechanism and the accuracy of the judgment of latent defects. It also elevates the multispectral detection results from simple anomaly identification to structural defect judgment supported by energy evolution logic.

[0016] 3. By determining the current defect stage based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, and dynamically switching the decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule, the multispectral detection process can be transformed from the traditional "fixed weight fusion" mode to a dynamic decision-making mode of "stage adaptive dominance". Specifically, the invention first extracts the conflict confidence index, energy closed-loop integrity index, and risk budget balance for the current detection cycle. Based on the risk budget consumption corresponding to these three factors, a weighted calculation is performed to obtain the comprehensive defect status index. This index reflects the comprehensive degree of defect development in the target high-voltage equipment under a unified indicator system. Subsequently, the comprehensive defect status index is compared with a preset defect stage judgment threshold range to determine the defect stage of the target high-voltage equipment, enabling staged management of the defect identification process according to the defect evolution process. Based on this, according to a preset stage-dominant spectral mapping rule, the dominant spectrum corresponding to the current defect stage is searched in the stage-dominant spectral mapping rule library. This allows the decision-making process to be dominated by the most sensitive spectral information for different defect stages, thereby improving the utilization efficiency of abnormal information. Furthermore, by judging the consistency between the current dominant spectrum and the dominant spectrum of the previous detection cycle, and combining this with stage stability judgment conditions to control the switching timing of the dominant spectrum, the problem of frequent spectral switching caused by short-term fluctuations or occasional interference can be effectively avoided. Through the above mechanism, this invention not only achieves adaptive weight allocation of multispectral information at different defect stages but also improves the stability and targeting of detection decisions, thereby significantly improving the accuracy and reliability of identifying latent defects in high-voltage equipment. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the specific steps of an embodiment of a multispectral fusion method for detecting latent defects in high-voltage equipment according to the present invention.

[0018] Figure 2 This is a schematic diagram of the module connection of an embodiment of a multispectral fusion high-voltage equipment latent defect detection system according to the present invention. Detailed Implementation

[0019] The following examples and... Figures 1-2 The present invention will be described in further detail, but the embodiments of the present invention are not limited thereto.

[0020] This invention discloses a method for detecting latent defects in high-voltage equipment using multispectral fusion, specifically including the following steps: Step S1: Collect multispectral data of the target high-voltage equipment within the same detection time window, and perform time alignment and spatial mapping processing on the multispectral data to form a unified spectral feature set; Step S2: Construct a spectral anomaly conflict matrix, perform cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identify conflict types, and calculate the conflict confidence index; Step S3: Based on the conflict type, construct a defect energy closed-loop verification model, verify the closed-loop integrity of the energy release, energy transfer and energy deposition processes of the corresponding abnormal features, and calculate the energy closed-loop integrity index. Step S4: Set a risk budget limit for the target high-voltage equipment, calculate the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index, and obtain the risk budget balance. Step S5: Based on the conflict credibility index, energy closed-loop integrity index and risk budget balance, determine the current defect stage, and switch the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier; Step S6: When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, it is determined that the target high-voltage equipment has a structural hidden defect, and the corresponding defect type and risk level are output.

[0021] In practical applications, multispectral data of the target high-voltage equipment is acquired within the same detection time window. Time alignment and spatial mapping of different spectral data are then performed to eliminate differences in sampling time and detection location between different sensing methods. This achieves a unified expression of multi-source detection information in both time and space dimensions, thereby constructing a unified spectral feature set. This ensures comparability and correlation between different spectral features, improving the reliability and consistency of multispectral data fusion analysis. Furthermore, by constructing a spectral anomaly conflict matrix, cross-spectral correlation analysis is performed on anomaly features in the unified spectral feature set. Anomaly conflict types are identified based on the anomaly consistency, differences, and response relationships between different spectral signals. The conflict confidence index is calculated through the conflict relationship strength, thereby identifying whether there is a structural correlation between different spectral anomalies. This distinguishes between multi-source anomalies caused by real defects and environmental noise or single-sensor anomalies, improving the credibility and reliability of anomaly information during defect identification. Based on the identified conflict types, a defect energy closed-loop verification model is constructed. By analyzing the energy release, energy transfer, and energy deposition processes corresponding to abnormal characteristics, a defect energy evolution path is established, and the closed-loop integrity of this path is verified. An energy closed-loop integrity index is calculated, enabling verification of abnormal signals from a physical mechanism perspective. This determines whether the anomaly conforms to the energy propagation law of equipment defect evolution, thereby further eliminating abnormal signals caused by non-defect factors and improving the physical rationality and accuracy of latent defect identification. A risk budget management mechanism is established for the target high-voltage equipment. By setting a risk budget for the equipment and calculating the risk consumption value corresponding to abnormal events based on the conflict credibility index and the energy closed-loop integrity index, the current risk budget balance of the equipment is obtained. The detection results are transformed into quantifiable risk indicators, enabling dynamic assessment and management of equipment operation risks and providing a quantitative basis for subsequent defect stage judgment. Based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, a comprehensive assessment of the current defect development status of the equipment is conducted to determine the defect stage of the equipment. The dominant spectrum for current detection and analysis is dynamically switched according to a preset stage-dominant spectral mapping rule, and a dominance identifier is generated. This allows for the selection of more representative spectral information as the basis for decision-making based on different defect stages, thereby improving the information utilization efficiency and judgment accuracy in multispectral fusion detection and enabling adaptive adjustment of the detection strategy. Under the condition of satisfying multiple conditions such as conflict credibility verification, energy closed-loop integrity verification, and defect stage stability verification, a final determination is made as to whether the target high-voltage equipment has structural latent defects. The specific defect type is identified by combining dominant spectral characteristics, and the equipment risk level is assessed based on the risk budget balance. This achieves comprehensive identification and risk assessment of latent defects in high-voltage equipment, improves the accuracy and reliability of defect detection results, and provides an effective basis for equipment operation and maintenance decisions.

[0022] The steps of acquiring multispectral data of the target high-voltage equipment within the same detection time window, and performing time alignment and spatial mapping processing on the multispectral data to form a unified spectral feature set are as follows: Step S11: Collect multispectral data of the target high-voltage equipment in the same detection time window. The multispectral data includes infrared radiation data, ultraviolet radiation data and visible light data. Collect the timestamp information corresponding to each spectral data. Step S12: Calculate the time offset between each spectral data based on the timestamp information, and perform time compensation processing on each spectral data according to the time offset so that each spectral data corresponds to the same unified time axis. Step S13: Select the image corresponding to the visible light data as the reference coordinate system, perform spatial feature matching on the infrared radiation data and ultraviolet radiation data, calculate the spatial transformation matrix, and map each spectral data to a unified spatial coordinate system. Step S14: Based on the unified time axis and unified spatial coordinate system, integrate the intensity features and time variation features corresponding to each spectral data to construct a unified spectral feature set.

[0023] In practical applications, by performing time alignment and spatial mapping on infrared, ultraviolet, and visible light data collected from the target high-voltage equipment within the same detection time window, the differences in sampling time and detection location of different spectral data are eliminated. This allows observation information from different spectral channels to be expressed and correlated under a unified time axis and a unified spatial coordinate system. Specifically, time offset calculation and time compensation processing are used to synchronize the spectral data, ensuring that different spectral signals have a consistent reference standard in the time dimension. A spatial transformation matrix is ​​established using the visible light image as a reference coordinate system to achieve spatial mapping of infrared and ultraviolet radiation data at the equipment's structural location, thus ensuring that different spectral anomaly information can be correlated and compared at the same spatial location. Based on this, the intensity characteristics and temporal variation characteristics of each spectral data are fused and integrated to construct a unified spectral feature set. This provides a unified and reliable data foundation for subsequent cross-spectral anomaly relationship analysis, conflict identification, and defect mechanism verification, thereby improving the accuracy and consistency of multispectral fusion analysis.

[0024] The steps of constructing a spectral anomaly conflict matrix, performing cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identifying conflict types, and calculating conflict confidence indices are as follows: Step S21: Extract the anomalous features corresponding to each spectrum from the unified spectral feature set. The anomalous features include intensity anomalous features, temporal abrupt change features, and spatial anomalous region features. Standardize the anomalous features to form an anomalous feature vector set. Step S22: Combine the abnormal feature vector sets in pairs, calculate the abnormal intensity difference value, response time difference and spatial overlap degree respectively, and construct the spectral abnormality conflict matrix. Step S23: Match each matrix element in the spectral anomaly conflict matrix with a preset conflict type rule base to identify the corresponding conflict type. The conflict types include intensity reverse conflict type, time lag anomaly type, spatial separation type and cooperative consistency type. Step S24: Determine whether there is a corresponding preset interpretation path for the conflict type. If there is an interpretation path, mark it as a structural conflict. Step S25: Calculate the conflict weight based on the significance of the abnormal intensity and spatial overlap of the structural conflict, and perform a weighted calculation on all the structural conflicts to obtain the conflict credibility index.

[0025] In practical applications, a spectral anomaly conflict matrix is ​​constructed to perform cross-spectral relationship analysis on anomalous features in a unified spectral feature set, thereby identifying structural correlation features between different spectral anomalies and calculating a conflict confidence index. Specifically, firstly, intensity anomaly features, temporal abrupt change features, and spatial anomaly region features corresponding to each spectrum are extracted from the unified spectral feature set, and these are standardized to form an anomaly feature vector set to ensure comparability of different spectral features at the same scale. Subsequently, the anomaly feature vectors are combined pairwise to calculate the anomaly intensity difference, response time difference, and spatial overlap, constructing a spectral anomaly conflict matrix so that the correlation between different spectral anomalies can be expressed in a structured form. Based on this, the conflict type is identified by matching with a preset conflict type rule base, and it is further determined whether there is a reasonable explanatory path for the conflict relationship, thereby filtering out structural conflicts with physical or mechanistic significance. Finally, the conflict weights are calculated based on the significance of the structural conflict anomaly intensity and the degree of spatial overlap. The conflict credibility index is obtained through weighted calculation, which can effectively distinguish between multispectral collaborative anomalies caused by real defects and pseudo-anomalies caused by environmental interference or single sensor anomalies. This improves the credibility of the anomaly identification results and provides a reliable basis for subsequent defect mechanism verification and risk assessment.

[0026] The steps for calculating conflict weights based on the significance of the anomalous intensity and spatial overlap of the structural conflicts, and then weighting all the structural conflicts to obtain the conflict credibility index, are as follows: Step S251: Extract conflict units marked as structural conflicts from the spectral anomaly conflict matrix to form a structural conflict set; Step S252: For each structural conflict in the set of structural conflicts, obtain the abnormal intensity value of the corresponding spectrum, calculate its deviation from the standard value, and standardize the deviation to obtain the significance of the abnormal intensity. Step S253: For each structural conflict in the set of structural conflicts, obtain the set of spatial coordinates of the corresponding spectral anomaly region, calculate the cross-union ratio between the anomaly regions, and obtain the spatial overlap. Step S254: Based on the significance of the anomaly intensity and the spatial overlap, calculate the conflict weights according to a preset ratio, and sum all the conflict weights in the structural conflict set to obtain the conflict credibility index.

[0027] In practical applications, by quantitatively evaluating the structural conflicts identified in the spectral anomaly conflict matrix, two feature indicators—anomaly intensity significance and spatial overlap—are introduced. The importance of different conflict units is weighted and modeled, and a unified conflict credibility index is formed through weighted calculation. This enables a credible quantitative evaluation of the structural contradictions between multi-source spectral anomalies. It can effectively distinguish between occasional anomalies caused by random noise and local disturbances and stable conflict modes caused by internal structural degradation of equipment. This provides a reliable quantitative basis for subsequent determination of structural latent defects, improving the accuracy and stability of defect identification results.

[0028] Based on the aforementioned conflict types, a defect energy closed-loop verification model is constructed. The steps for verifying the closed-loop integrity of the energy release, energy transfer, and energy deposition processes for the corresponding anomaly characteristics, and calculating the energy closed-loop integrity index, are as follows: Step S31: According to the conflict type, call the corresponding energy evolution template, extract abnormal features from the unified spectral feature set, and calculate the energy release feature parameters, energy transfer feature parameters, and energy deposition feature parameters respectively; Step S32: Based on the release time in the energy release characteristic parameters, the transfer response time in the energy transfer characteristic parameters, and the precipitation formation time in the energy precipitation characteristic parameters, perform time sequence verification and calculate the time continuity parameter of energy transfer; Step S33: Based on the release location in the energy release characteristic parameters, the transfer path information in the energy transfer characteristic parameters, and the deposition location in the energy deposition characteristic parameters, perform spatial relationship analysis and calculate the spatial consistency parameter of energy propagation. Step S34: Based on the release intensity in the energy release characteristic parameters, the transfer intensity attenuation ratio in the energy transfer characteristic parameters, and the cumulative value of the precipitation intensity in the energy precipitation characteristic parameters, calculate the attenuation rationality parameter of energy attenuation. Step S35: Normalize the time continuity parameter, spatial consistency parameter, and attenuation rationality parameter, and perform weighted calculation according to preset weights to obtain the energy closed-loop integrity index.

[0029] In practical applications, based on the identification of structural conflict types, energy evolution mechanisms are introduced to verify the causal closed-loop of anomalous features. By constructing a defect energy closed-loop verification model, the physical processes corresponding to spectral anomalies are abstracted into three continuous stages: energy release, energy transfer, and energy deposition. The logical consistency between anomalous features is systematically verified from three dimensions: temporal sequence, spatial propagation relationship, and energy attenuation law. By calculating temporal continuity parameters, spatial consistency parameters, and attenuation rationality parameters, and further fusing them to obtain the energy closed-loop integrity index, it is determined whether the anomalous signal can form a complete and reasonable energy evolution chain. This effectively eliminates non-structural anomalous signals caused by random interference, environmental noise, or sporadic anomalies, ensuring that only anomalous patterns conforming to physical evolution laws are used for subsequent defect determination. This improves the reliability and scientific rigor of latent defect identification and provides a stable and reliable basis for structural defect risk assessment.

[0030] The steps for setting a risk budget for the target high-voltage equipment, calculating the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index, and obtaining the remaining risk budget are as follows: Step S41: Obtain the equipment operation information of the target high-voltage equipment, and calculate the risk budget amount of the target high-voltage equipment in the current operating cycle according to the preset risk budget model; Step S42: Extract the conflict confidence index and energy closed-loop integrity index within the current detection time window, and establish corresponding abnormal event records; Step S43: Normalize the conflict credibility index and calculate the conflict risk consumption value by combining it with the preset conflict type weight coefficient. Step S44: Normalize the energy closed-loop integrity index and calculate the energy closed-loop risk consumption value by combining it with the preset energy evolution weight coefficient. Step S45: The conflict risk consumption value and the energy closed-loop risk consumption value are weighted according to the preset risk fusion weight to obtain the comprehensive risk consumption value. Step S46: The comprehensive risk consumption value is added to the cumulative risk consumption value of the target high-voltage equipment for updating. Based on the risk budget amount and the cumulative risk consumption value, the risk budget balance of the target high-voltage equipment is calculated.

[0031] In practical applications, by establishing a risk budget mechanism for target high-voltage equipment, abnormal information during defect identification is transformed into quantifiable and cumulative risk consumption indicators, thereby achieving dynamic management of equipment operation risks. Specifically, by acquiring equipment operation information and combining it with a preset risk budget model, the risk budget that the equipment can withstand within the current operating cycle is determined. Then, the conflict credibility index and energy closed-loop integrity index are mapped to corresponding risk consumption values, and fused according to preset weights to obtain a comprehensive risk consumption value. By continuously accumulating the comprehensive risk consumption value and comparing it with the risk budget, the risk budget balance of the equipment is calculated in real time, enabling the system to reflect the current remaining risk-bearing capacity of the equipment. This not only unifies the scattered abnormal detection results into quantitative risk indicators but also reflects the development trend of latent defects through the dynamic changes in the budget balance, thus providing a reliable basis for subsequent defect stage judgment, risk warning, and operation and maintenance decisions, and improving the continuity and foresight of latent defect management of power transmission and transformation equipment.

[0032] The steps for determining the current defect stage based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, and for switching the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominance identifier are as follows: Step S51: Extract the conflict credibility index, energy closed-loop integrity index and risk budget balance for the current detection cycle, and perform a weighted calculation on the risk budget consumption corresponding to the conflict credibility index, energy closed-loop integrity index and risk budget balance to obtain the defect comprehensive status index. Step S52: Compare the comprehensive defect status index with the preset defect stage judgment threshold range to determine the current defect stage of the target high-voltage equipment; Step S53: According to the preset stage-dominant spectrum mapping rule, search for the dominant spectrum corresponding to the current defect stage in the stage-dominant spectrum mapping rule library; Step S54: Determine whether the dominant spectrum is consistent with the dominant spectrum of the previous detection cycle. If the dominant spectrum is inconsistent with the dominant spectrum of the previous detection cycle, determine whether to perform dominant spectrum switching according to the preset stage stability judgment condition, and generate a dominant power identifier according to the determined dominant spectrum.

[0033] In practical applications, a comprehensive defect status index is constructed by integrating and analyzing the conflict credibility index, energy closed-loop integrity index, and risk budget balance. This unifies the mapping of multi-source anomaly characteristics, energy evolution characteristics, and risk consumption characteristics into a comprehensive status indicator that characterizes the degree of equipment defect evolution. By matching this comprehensive status index with a preset defect stage judgment threshold range, the current defect stage of the target high-voltage equipment can be clearly defined, realizing the transformation from single anomaly judgment to staged defect identification. Furthermore, by introducing a stage-dominant spectrum mapping rule, different defect stages correspond to different decision-making dominant spectra. This allows for dynamic adjustment of the dominant analysis dimension during multispectral information fusion analysis, enabling the system to prioritize judgment based on the spectral information that best reflects the current defect mechanism, improving the targeting and accuracy of defect identification. Simultaneously, by setting dominant spectrum consistency judgment and stage stability judgment conditions, frequent switching of dominant spectra due to short-term fluctuations is avoided, ensuring the stability and continuity of the analysis process. The final generated dominant authority identifier provides a unified decision-making basis for subsequent defect type judgment, risk level assessment, and early warning decisions, thereby realizing staged management and adaptive decision control in the process of identifying latent defects in power transmission and transformation equipment.

[0034] The steps for determining that the target high-voltage equipment has a structural latent defect and outputting the corresponding defect type and risk level are as follows: when the conflict credibility index meets the first threshold condition, the energy closed-loop integrity index meets the second threshold condition, and the dominance identifier meets the phase consistency condition. Step S61: Extract the conflict confidence index, energy closed-loop integrity index and dominance identifier for the current detection cycle, and extract the current dominant spectral type and corresponding defect stage information from the dominance identifier; Step S62: Compare the conflict credibility index with a preset credibility threshold. When the conflict credibility index is greater than or equal to the preset credibility threshold, determine that the abnormal conflict structure meets the credibility condition. Step S63: Compare the energy closed-loop integrity index with a preset integrity threshold. When the energy closed-loop integrity index is greater than or equal to the preset integrity threshold, it is determined that the energy closed-loop integrity condition is abnormally met. Step S64: Based on the defect stage information in the dominance identifier, perform consistency verification on the stage information of the current detection cycle and the historical detection cycle. When the stage is continuously and stably reached a preset number of cycles, it is determined that the stage consistency condition is met. Step S65: When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, it is determined that the target high-voltage equipment has a structural hidden defect. Step S66: Call the preset defect feature library according to the dominant spectral type, and match and identify the current spectral anomaly features to determine the corresponding defect type; Step S67: Based on the conflict credibility index, energy closed-loop integrity index and risk budget balance, a comprehensive evaluation is performed to determine the risk level of the target high-voltage equipment. Combining the defect type and risk level, the defect detection result is output.

[0035] In practical applications, a multi-condition collaborative identification mechanism for structural latent defects is constructed by jointly judging the conflict credibility index, energy closed-loop integrity index, and dominance identifier, thereby avoiding misjudgments caused by single anomaly features. First, by comparing the conflict credibility index with a preset credibility threshold, it can be determined whether a credible conflict structure is formed between multispectral anomalies, thus filtering out anomalies with structural correlation characteristics and avoiding false anomalies caused by single spectral noise or random disturbances being misidentified as equipment defects. Second, by verifying the energy closed-loop integrity index with a preset integrity threshold, it can be further verified whether the anomaly conforms to the basic laws of defect energy evolution, that is, whether the anomaly can form a reasonable and complete energy evolution closed loop in the three stages of energy release, energy transfer, and energy deposition, thus ensuring the physical rationality of the identified anomaly. Based on this, by verifying the historical cycle consistency of the defect stage information in the dominance identifier, it can be determined whether the equipment defect evolution maintains stable stage characteristics over multiple detection cycles, thereby avoiding interference from short-term fluctuations or occasional anomalies on the defect identification results and improving the stability and reliability of defect identification. By simultaneously satisfying the conflict credibility condition, the energy closed-loop integrity condition, and the stage consistency condition, the system can comprehensively determine whether abnormal features constitute structural latent defects with continuously evolving characteristics, thus realizing the transformation from anomaly detection to defect mechanism identification. Furthermore, by calling a preset defect feature library based on the dominant spectral type and matching and identifying the current spectral anomaly features, the corresponding defect type can be determined, allowing the defect identification results to be directly linked to the specific equipment defect mechanism. Finally, by combining the conflict credibility index, the energy closed-loop integrity index, and the risk budget balance for comprehensive evaluation, the potential operational risks posed by defects can be quantitatively assessed, the corresponding risk level determined, and this, together with the defect type, constitutes a complete defect detection result. This not only enables accurate identification of structural latent defects in power transmission and transformation equipment but also simultaneously outputs the defect type and risk level, providing a reliable basis for subsequent equipment operation and maintenance decisions, maintenance prioritization, and risk warnings.

[0036] A multispectral fusion-based system for detecting latent defects in high-voltage equipment, comprising the following steps, using the multispectral fusion-based method described above: The spectral feature fusion module collects multispectral data of the target high-voltage equipment within the same detection time window, performs time alignment and spatial mapping processing on the multispectral data, and forms a unified set of spectral features. The spectral conflict analysis module constructs a spectral anomaly conflict matrix, performs cross-spectral conflict relationship analysis on the anomalous features in the unified spectral feature set, identifies conflict types, and calculates the conflict confidence index. The energy closed-loop verification module, based on the conflict type, constructs a defective energy closed-loop verification model, verifies the closed-loop integrity of the energy release, energy transfer and energy deposition processes of the corresponding abnormal features, and calculates the energy closed-loop integrity index. The risk budget assessment module sets a risk budget limit for the target high-voltage equipment, calculates the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index, and obtains the risk budget balance. The stage determination and spectrum switching module determines the current defect stage based on the conflict credibility index, energy closed-loop integrity index and risk budget balance, and switches the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier. The defect result output module determines that the target high-voltage equipment has a structural latent defect when the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, and outputs the corresponding defect type and risk level.

[0037] In practical applications, a spectral feature fusion module performs time alignment and spatial mapping on the multispectral detection data of the target high-voltage equipment, achieving a unified expression of different spectral information and providing a consistent feature foundation for subsequent cross-spectral anomaly correlation analysis. A spectral conflict analysis module constructs a spectral anomaly conflict matrix to identify conflict relationships between different spectral anomalies and calculates a conflict credibility index to uncover structural correlations between multispectral anomalies that may reflect equipment defect mechanisms. An energy closed-loop verification module establishes a defect energy evolution closed-loop model based on conflict types to verify the integrity of anomaly features during energy release, transfer, and deposition, confirming the rationality of the anomalies from a mechanistic perspective. A risk budget assessment module uses a risk budget mechanism to quantify the risk of the conflict credibility index and the energy closed-loop integrity index, calculating the equipment's risk budget balance within the current operating cycle to reflect the impact of potential defects on equipment operation risks. A stage determination and spectral switching module determines the evolution stage of equipment defects based on anomaly credibility, energy evolution integrity, and risk budget status, and dynamically switches the current decision-making dominant spectrum to improve the targeting of defect identification at different stages. The defect result output module determines that the equipment has structural hidden defects when the conditions of credibility, completeness and phase consistency are met, and outputs the corresponding defect type and risk level to provide a basis for equipment operation and maintenance decisions.

[0038] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for detecting latent defects in high-voltage equipment using multispectral fusion, characterized in that, Includes the following steps: Multispectral data of the target high-voltage equipment are collected within the same detection time window. The multispectral data is then processed by time alignment and spatial mapping to form a unified set of spectral features. Construct a spectral anomaly conflict matrix, perform cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identify conflict types and calculate conflict confidence index; Based on the aforementioned conflict types, a defect energy closed-loop verification model is constructed to verify the closed-loop integrity of the energy release, energy transfer, and energy deposition processes for the corresponding abnormal features, and to calculate the energy closed-loop integrity index. Set a risk budget limit for the target high-voltage equipment, and calculate the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index to obtain the risk budget balance. Based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, the current defect stage is determined, and the current decision-making dominant spectrum is switched according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier; When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the phase consistency condition, it is determined that the target high-voltage equipment has a structural latent defect, and the corresponding defect type and risk level are output.

2. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The steps of collecting multispectral data from the target high-voltage equipment within the same detection time window, and performing time alignment and spatial mapping processing on the multispectral data to form a unified spectral feature set are as follows: Multispectral data of the target high-voltage equipment are collected within the same detection time window. The multispectral data includes infrared radiation data, ultraviolet radiation data, and visible light data. The timestamp information corresponding to each spectral data is collected respectively. The time offset between each spectral data is calculated based on the timestamp information, and time compensation processing is performed on each spectral data according to the time offset to make each spectral data correspond to the same unified time axis. The image corresponding to the visible light data is selected as the reference coordinate system. Spatial feature matching is performed on the infrared radiation data and ultraviolet radiation data, and the spatial transformation matrix is ​​calculated to map each spectral data to a unified spatial coordinate system. Based on the unified time axis and unified spatial coordinate system, the intensity features and time variation features corresponding to each spectral data are integrated to construct a unified spectral feature set.

3. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The steps of constructing the spectral anomaly conflict matrix, performing cross-spectral conflict relationship analysis on the anomaly features in the unified spectral feature set, identifying conflict types, and calculating the conflict confidence index are as follows: The anomalous features corresponding to each spectrum are extracted from the unified spectral feature set. The anomalous features include intensity anomalous features, temporal abrupt change features, and spatial anomalous region features. The anomalous features are then standardized to form an anomalous feature vector set. The abnormal feature vector set is combined in pairs, and the abnormal intensity difference value, response time difference and spatial overlap are calculated respectively to construct the spectral abnormality conflict matrix. Each matrix element in the spectral anomaly conflict matrix is ​​matched with a preset conflict type rule base to identify the corresponding conflict type. The conflict types include intensity-reverse conflict type, time-lag anomaly type, spatial separation type, and cooperative consistency type. Determine whether the conflict type has a corresponding preset interpretation path. If an interpretation path exists, mark it as a structural conflict. Based on the significance of the abnormal intensity and spatial overlap of the structural conflicts, the conflict weights are calculated, and all the structural conflicts are weighted to obtain the conflict credibility index.

4. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 3, characterized in that, The step of calculating conflict weights based on the significance of the anomaly intensity and spatial overlap of the structural conflicts, and then performing a weighted calculation on all the structural conflicts to obtain a conflict credibility index, specifically includes: The conflict units marked as structural conflicts are extracted from the spectral anomaly conflict matrix to form a structural conflict set; For each structural conflict in the set of structural conflicts, the abnormal intensity value of the corresponding spectrum is obtained, the degree of deviation from the standard value is calculated, and the degree of deviation is standardized to obtain the significance of the abnormal intensity. For each structural conflict in the set of structural conflicts, obtain the set of spatial coordinates of the corresponding spectral anomaly region, calculate the cross-union ratio between the anomaly regions, and obtain the spatial overlap. Based on the significance of the anomaly intensity and the degree of spatial overlap, the conflict weights are calculated according to a preset ratio, and the conflict weights of all conflicts in the structural conflict set are summed to obtain the conflict credibility index.

5. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The steps of constructing a defect energy closed-loop verification model based on the conflict type, verifying the closed-loop integrity of the energy release, energy transfer, and energy deposition processes for the corresponding abnormal features, and calculating the energy closed-loop integrity index are as follows: According to the conflict type, the corresponding energy evolution template is invoked, abnormal features are extracted from the unified spectral feature set, and energy release feature parameters, energy transfer feature parameters and energy deposition feature parameters are calculated respectively. Based on the release time in the energy release characteristic parameters, the transfer response time in the energy transfer characteristic parameters, and the precipitation formation time in the energy precipitation characteristic parameters, the time sequence is verified, and the time continuity parameter of energy transfer is calculated. Based on the release location in the energy release characteristic parameters, the transfer path information in the energy transfer characteristic parameters, and the deposition location in the energy deposition characteristic parameters, a spatial relationship analysis is performed to calculate the spatial consistency parameter of energy propagation. Based on the release intensity in the energy release characteristic parameters, the transfer intensity attenuation ratio in the energy transfer characteristic parameters, and the cumulative value of the precipitation intensity in the energy precipitation characteristic parameters, the attenuation rationality parameter of energy attenuation is calculated. The time continuity parameter, spatial consistency parameter, and attenuation rationality parameter are normalized and weighted according to preset weights to obtain the energy closed-loop integrity index.

6. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The step of setting a risk budget for the target high-voltage equipment and calculating the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index to obtain the risk budget balance is as follows: Obtain the equipment operation information of the target high-voltage equipment, and calculate the risk budget amount of the target high-voltage equipment in the current operating cycle according to the preset risk budget model; Extract the conflict confidence index and energy closed-loop integrity index within the current detection time window, and establish corresponding abnormal event records; The conflict credibility index is normalized and combined with a preset conflict type weighting coefficient to calculate the conflict risk consumption value. The energy closed-loop integrity index is normalized and combined with a preset energy evolution weighting coefficient to calculate the energy closed-loop risk consumption value. The conflict risk consumption value and the energy closed-loop risk consumption value are weighted and calculated according to the preset risk fusion weight to obtain the comprehensive risk consumption value. The comprehensive risk consumption value is added to the cumulative risk consumption value of the target high-voltage equipment for updating, and the risk budget balance of the target high-voltage equipment is calculated based on the risk budget amount and the cumulative risk consumption value.

7. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The step of determining the current defect stage based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, and switching the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominance identifier is as follows: Extract the conflict confidence index, energy closed-loop integrity index, and risk budget balance for the current detection cycle, and perform a weighted calculation on the risk budget consumption corresponding to the conflict confidence index, energy closed-loop integrity index, and risk budget balance to obtain the defect comprehensive status index. The current defect stage of the target high-voltage equipment is determined by comparing the comprehensive defect status index with a preset defect stage judgment threshold range. According to the preset stage-dominant spectrum mapping rules, the dominant spectrum corresponding to the current defect stage is searched in the stage-dominant spectrum mapping rule library; It is determined whether the dominant spectrum is consistent with the dominant spectrum of the previous detection cycle. If the dominant spectrum is inconsistent with the dominant spectrum of the previous detection cycle, it is determined whether to perform a dominant spectrum switch according to the preset stage stability judgment condition, and a dominant power identifier is generated according to the determined dominant spectrum.

8. The method for detecting latent defects in high-voltage equipment using multispectral fusion according to claim 1, characterized in that, The step of determining that the target high-voltage equipment has a structural latent defect and outputting the corresponding defect type and risk level when the conflict credibility index meets the first threshold condition, the energy closed-loop integrity index meets the second threshold condition, and the dominance identifier meets the phase consistency condition is as follows: Extract the conflict confidence index, energy closed-loop integrity index and dominance identifier of the current detection cycle, and extract the current dominant spectral type and corresponding defect stage information from the dominance identifier; The conflict credibility index is compared with a preset credibility threshold. When the conflict credibility index is greater than or equal to the preset credibility threshold, the abnormal conflict structure is determined to meet the credibility condition. The energy closed-loop integrity index is compared with a preset integrity threshold. When the energy closed-loop integrity index is greater than or equal to the preset integrity threshold, it is determined that the energy closed-loop integrity condition is abnormally met. Based on the defect stage information in the dominant identifier, the consistency of the stage information of the current detection cycle and the historical detection cycle is checked. When the stage is continuously and stably reached a preset number of cycles, it is determined that the stage consistency condition is met. When the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, it is determined that the target high-voltage equipment has a structural hidden defect. The preset defect feature library is invoked according to the dominant spectral type, and the current spectral anomaly features are matched and identified to determine the corresponding defect type. Based on the conflict credibility index, energy closed-loop integrity index, and risk budget balance, a comprehensive assessment is conducted to determine the risk level of the target high-voltage equipment. Combining the defect type and risk level, the defect detection results are output.

9. A multispectral fusion system for detecting latent defects in high-voltage equipment, characterized in that, The method for detecting latent defects in high-voltage equipment using multispectral fusion as described in any one of claims 1-8 includes: The spectral feature fusion module collects multispectral data of the target high-voltage equipment within the same detection time window, performs time alignment and spatial mapping processing on the multispectral data, and forms a unified set of spectral features. The spectral conflict analysis module constructs a spectral anomaly conflict matrix, performs cross-spectral conflict relationship analysis on the anomalous features in the unified spectral feature set, identifies conflict types, and calculates the conflict confidence index. The energy closed-loop verification module, based on the conflict type, constructs a defective energy closed-loop verification model, verifies the closed-loop integrity of the energy release, energy transfer and energy deposition processes of the corresponding abnormal features, and calculates the energy closed-loop integrity index. The risk budget assessment module sets a risk budget limit for the target high-voltage equipment, calculates the corresponding risk consumption value based on the conflict credibility index and the energy closed-loop integrity index, and obtains the risk budget balance. The stage determination and spectrum switching module determines the current defect stage based on the conflict credibility index, energy closed-loop integrity index and risk budget balance, and switches the current decision-making dominant spectrum according to the preset stage-dominant spectrum mapping rule to generate a dominant power identifier. The defect result output module determines that the target high-voltage equipment has a structural latent defect when the conflict credibility index meets the credibility threshold condition, the energy closed-loop integrity index meets the integrity threshold condition, and the dominance identifier meets the stage consistency condition, and outputs the corresponding defect type and risk level.

Images