Distribution box switch cabinet fault detection method and system

Abstract

The invention belongs to the technical field of power fault detection, and particularly relates to a distribution box switch cabinet fault detection method and system, and the method comprises the steps: obtaining the original temperature matrix data of a distribution box switch cabinet, carrying out the binary segmentation of the original temperature matrix data, and extracting all connected regions as candidate hot spot regions; polar coordinate sampling is carried out on each candidate hot spot area, and thermal field regularity is calculated; performing morphological processing on each candidate hot spot area, determining and counting the total enthalpy amount in the body and the external overflow heat energy of the peripheral annular area, and calculating an edge heat capacity transition coefficient; and performing weighted fusion on the thermal field regularity and the edge thermal capacity transition coefficient, calculating a decision score, and when the decision score is greater than a preset decision threshold, determining that the candidate hot spot region is a real thermal fault. According to the method, accurate identification of the real thermal fault and the reflection interference can be realized based on the morphological characteristics of the thermal physical field.

Description

Technical Field

[0001] This invention relates to the field of power fault detection technology. More specifically, this invention relates to a method and system for detecting faults in distribution box switchgear. Background Technology

[0002] As a critical terminal node in the power system, the distribution box switchgear contains a dense array of busbar connection points, circuit breaker contacts, and various cable joints. During long-term operation under high current, these components are prone to increased contact resistance due to loosening caused by mechanical vibration or oxidation corrosion from environmental factors, leading to localized overheating.

[0003] Currently, infrared thermal imaging technology, with its advantages of non-contact measurement and visual imaging, has become the main means of detecting thermal faults in such power equipment. Most existing infrared detection algorithms are still in the rudimentary stage based on pixel-level statistics. The common practice is to simply set a fixed temperature threshold and judge any pixel exceeding the temperature threshold as faulty.

[0004] However, the internal environment of switchgear is complex, with a large number of highly reflective metal surfaces, such as shiny copper busbars, silver-plated contacts, and stainless steel components. These highly reflective metal surfaces have extremely low infrared emissivity, which easily reflects the heat radiation from external heat sources such as transformer casings, lighting fixtures, and even the heating elements of the inspection robot itself into the thermal imager lens of the inspection robot. This makes it impossible for the inspection robot to effectively distinguish between real overheating originating from inside the equipment and false heat signals caused by reflection, resulting in a high false alarm rate. Maintenance personnel have to perform a lot of ineffective and tedious manual verification, which seriously limits the further improvement of the intelligent operation and maintenance level of the power distribution network. Summary of the Invention

[0005] To address the high false alarm rate of existing infrared detection technologies due to metal reflection interference, this invention proposes a fault detection method and system for distribution boxes and switchgear, which can accurately identify real thermal faults and reflection interference based on the morphological characteristics of thermophysical fields.

[0006] In a first aspect, the present invention provides a fault detection method for a distribution box switchgear, comprising: acquiring raw temperature matrix data output by an infrared detector of the distribution box switchgear; calculating a full-image temperature histogram to determine an ambient reference temperature; setting an adaptive floating threshold based on the ambient reference temperature; using the adaptive floating threshold to perform binarization segmentation on the raw temperature matrix data; extracting all connected regions as candidate hotspot regions; performing polar coordinate sampling on each candidate hotspot region; drawing multiple rays outward from the highest temperature point in the region as the center; retrieving the temperature distribution and boundary distance on each ray; and determining the results based on the retrieval. The thermal field regularity is calculated, which characterizes the degree to which the regional heat distribution conforms to the physical heat source characteristics. Morphological processing is performed on each candidate hotspot region to determine its body and surrounding annular region. The total enthalpy within the body and the external overflow heat energy of the surrounding annular region are statistically analyzed. An edge thermal capacity transition coefficient, characterizing the energy cutoff at the region's edge, is calculated. The thermal field regularity and the edge thermal capacity transition coefficient are weighted and fused to calculate the final decision score. When the decision score is greater than a preset decision threshold, the candidate hotspot region is determined to be a real thermal fault; otherwise, it is determined to be an interference signal.

[0007] By adopting the above technical solution, the original temperature matrix data of the distribution box switchgear is obtained and binarized, solving the problem that existing infrared detection algorithms rely solely on pixel-level statistics and cannot distinguish between endogenous fault heat sources and exogenous reflected heat sources, resulting in a high false alarm rate. Based on the fact that real thermal faults follow the heat diffusion equation and that reflection interference depends on the physical characteristics of the geometric boundaries of the reflecting surface, this solution integrates thermal field regularity and edge thermal capacity transition coefficients to achieve accurate identification of faults and interference at the physical level. This effectively suppresses false thermal signal interference caused by the high reflectivity of the metal surface inside the switchgear, significantly improving the accuracy of thermal fault detection in power equipment.

[0008] Preferably, the step of setting an adaptive floating threshold based on the ambient reference temperature and using the adaptive floating threshold to perform binarization segmentation on the original temperature matrix data includes: extracting the temperature corresponding to the peak value of the histogram as the ambient reference temperature; setting a base temperature rise threshold and calculating the adaptive floating threshold, wherein the adaptive floating threshold is the sum of the ambient reference temperature and the base temperature rise threshold; marking pixels in the original temperature matrix data with temperature values ​​greater than the adaptive floating threshold as foreground and pixels with temperature values ​​less than or equal to the adaptive floating threshold as background, thereby extracting a set of connected regions.

[0009] By adopting the above technical solution, the temperature corresponding to the peak value of the histogram is extracted as the environmental reference temperature, and an adaptive floating threshold is set accordingly. The baseline of the binarized segmentation can be automatically adjusted according to the actual environmental temperature during detection, thereby effectively adapting to the temperature difference between winter and summer or the background temperature fluctuation of different substation sites. This ensures that the algorithm can accurately extract the set of connected regions containing real thermal faults and suspected interference under different environmental conditions, avoiding the problem of missed detection or incomplete region extraction caused by improper setting of fixed thresholds.

[0010] Preferably, the polar coordinate sampling for each candidate hotspot region includes: locating the highest temperature point within the candidate hotspot region as the center point; constructing a polar coordinate system with the center point as the origin, and drawing out a ray at every preset angle, for a total of multiple rays; searching along the direction of each ray until the boundary of the candidate hotspot region is encountered, and recording the Euclidean distance from the center point to the boundary and the actual temperature at the boundary.

[0011] By employing the aforementioned technical solution and using polar coordinate sampling, multiple rays are drawn outwards from the highest temperature point in the region, and the boundary distances are retrieved. This transforms the complex two-dimensional thermal field morphology into standardized radial distribution features. This not only eliminates the computational obstacles caused by the irregular shape of the hotspot region but also comprehensively captures the spatial structure information of the heat source spreading from the center to the edge from a geometric perspective. Establishing a unified polar coordinate system provides a rotationally invariant data foundation for subsequent calculations of thermal field regularity, ensuring that the feature extraction process is unaffected by the position and angle of the hotspot in the image.

[0012] Preferably, the thermal field regularity Satisfying the relation:

[0013] ; In the formula, Let the standard deviation of the temperature gradient of all rays be denoted as . It is the equilibrium constant. Temperature at the center point The actual temperature at the boundary. Let Euclidean distance be the distance from the center point to the boundary. For distance correction parameters, The total number of rays, Indicates the sequence number of the ray.

[0014] By employing the above technical solution, the regularity of the thermal field is calculated to characterize the degree to which the regional heat distribution conforms to the physical characteristics of the heat source. This solution assesses the uniformity of the thermal field by statistically analyzing the standard deviation of the temperature gradient in each ray direction. It can accurately identify asymmetric hot spots formed by reflections from irregular metal surfaces, thereby initially filtering out interference signals that do not conform to the physical laws of heat conduction during the feature extraction stage.

[0015] Preferably, the morphological processing of each candidate hotspot region includes: defining the candidate hotspot region as an ontology; and performing a morphological dilation operation on the ontology to generate an outer ring region with a preset width.

[0016] Preferably, the edge heat capacity transition coefficient Satisfying the relation: ; In the formula, Represents the entity, Represents the outer ring area. coordinates within the body The temperature at that location Coordinates within the outer ring area The temperature at that location The ambient reference temperature For positive correction parameters, As the base, This represents the pixel area of ​​the candidate hotspot region.

[0017] By employing the above technical solution, the constructed edge thermal capacity transition coefficient is calculated using a logarithmic function to determine the temperature difference ratio between the main body and the surrounding annular region. Utilizing the difference between a genuine thermal fault's gradual temperature transition due to thermal diffusion and a reflected interference signal's sharp temperature drop due to energy truncation, a highly sensitive discrimination index for reflected interference is established. This coefficient directly reflects the energy penetration at the edge of the target area, effectively distinguishing between genuine thermal faults with natural thermal diffusion characteristics and false signals with optical reflection characteristics.

[0018] Preferably, the decision scoring Satisfying the relation: ; In the formula, and These are the weighting coefficients for thermal field regularity and edge heat capacity transition coefficient, respectively. For thermal field regularity, This is a bias term.

[0019] By employing the aforementioned technical solution, a multi-dimensional model is constructed that comprehensively considers both the internal uniformity of the thermal field and the edge energy transition characteristics by weighted fusion of thermal field regularity and edge heat capacity transition coefficient to calculate the final decision score. By introducing bias terms and weighting coefficients, the system can balance the influence of different physical characteristics within a unified decision space, avoiding the potential bias of single-feature judgments and ensuring the rigor and reliability of the final judgment result in terms of physical logic.

[0020] Preferably, the weighting coefficient of the thermal field regularity is greater than the weighting coefficient of the edge thermal capacity transition coefficient, and the edge thermal capacity transition coefficient in the formula reflects the nonlinear penalty mechanism through the square term; when the decision score is greater than the preset decision threshold, the target is determined to be a real thermal fault and an alarm signal is generated.

[0021] Preferably, the calculation of the full-map temperature histogram to determine the environmental reference temperature includes: statistically analyzing the frequency of each temperature value in the original temperature matrix data to construct a temperature distribution histogram; smoothing the temperature distribution histogram to eliminate noise interference; identifying the position of the maximum peak in the smoothed histogram, and determining the horizontal axis temperature value corresponding to the maximum peak position as the environmental reference temperature to adapt to background temperature changes in different seasons.

[0022] Secondly, the present invention provides a fault detection system for distribution box switchgear, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned fault detection method for distribution box switchgear is implemented.

[0023] By adopting the above technical solution, a computer program for detecting faults in distribution boxes and switchgear is generated and stored in a memory, so that it can be loaded and executed by a processor. Terminal equipment can then be made based on the memory and processor for convenient use.

[0024] The beneficial effects of this invention are as follows: This invention goes beyond simple comparison of temperature values, and instead constructs a feature identification system based on the morphology of thermophysical fields. This scheme follows a processing logic from data to features, and then to decision-making. Based on the fundamental differences in physical manifestation between the heat conduction equation and the laws of geometric optics, it achieves accurate differentiation between endogenous fault heat sources and exogenous reflection interference.

[0025] Furthermore, this invention introduces a dynamic reference temperature, enabling the algorithm to adapt to significant temperature differences between winter and summer; and by introducing engineering constants to eliminate singularities in mathematical calculations, it ensures stable system operation under extreme conditions. In addition, this invention is built based on a physical model, eliminating the need for training with a large number of samples and possessing zero-sample cold-start capability, making it particularly suitable for practical applications where fault samples are scarce. Attached Figure Description

[0026] Figure 1 This is a flowchart of a fault detection method for distribution box switchgear according to the present invention; Figure 2 This is a schematic diagram comparing the radial temperature waveforms of the heat source and the interference signal in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the clustering and differentiation of faults and interference based on a two-dimensional feature space in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0028] This invention discloses a method for fault detection in distribution box switchgear, referring to... Figure 1 This includes steps S1-S4: S1. Obtain the original temperature matrix data output by the infrared detector of the distribution box switchgear, calculate the full-map temperature histogram to determine the ambient reference temperature, set an adaptive floating threshold based on the ambient reference temperature, and use the adaptive floating threshold to perform binarization segmentation on the original temperature matrix data to extract all connected regions as candidate hotspot regions.

[0029] In one optional embodiment, the system first connects to the infrared detector of the distribution box switchgear and reads the raw temperature matrix data output by the infrared detector in real time. Each element in the raw temperature matrix represents the absolute temperature at spatial coordinates. Since the ambient temperature varies significantly between winter and summer, this embodiment further processes the raw temperature matrix data to adapt to these temperature differences. Specifically, the system first calculates the temperature histogram of the entire image. After smoothing the histogram, the temperature corresponding to the peak value is determined as the ambient reference temperature. Then, a basic temperature rise threshold is set, and an adaptive floating threshold is calculated. In one specific implementation, the adaptive floating threshold is the ambient reference temperature plus the basic temperature rise threshold. Finally, the raw temperature matrix data is binarized using the adaptive floating threshold, and all connected regions with temperature values ​​greater than the adaptive floating threshold are extracted to form a candidate hotspot region set. It should be noted that the candidate hotspot region set obtained at this time includes both actual hotspots with poor contact and light spots formed by metal reflections.

[0030] To more clearly illustrate the process of constructing a candidate hotspot region set, the following embodiments of the present invention will be illustrated by examples: Suppose that in a certain test, the ambient baseline temperature obtained through histogram analysis is 25 degrees Celsius, and the set base temperature rise threshold is 10 degrees Celsius, then the adaptive floating threshold is 35 degrees Celsius, and all areas above 35 degrees Celsius will be extracted as candidate areas.

[0031] In this way, by constructing a dynamic benchmark and setting an adaptive threshold, the algorithm can be ensured to be unaffected by seasonal changes in ambient temperature, accurately identify all potential abnormally high temperature areas, and provide a complete candidate set for subsequent authenticity identification.

[0032] S2. Perform polar coordinate sampling on each candidate hotspot region, and draw multiple rays outward from the highest temperature point in the region. Search for the temperature distribution and boundary distance on each ray. Calculate the thermal field regularity based on the search results. The thermal field regularity is used to characterize the degree to which the regional heat distribution conforms to the physical heat source characteristics.

[0033] In an optional embodiment, within each candidate hotspot region, the highest temperature point within the region is first located as the center point. A polar coordinate system is then constructed with this center point as the origin. Simultaneously, a ray is drawn every 45 degrees, for a total of eight rays. The search continues along the direction of each ray until the boundary of the candidate hotspot region is encountered. The Euclidean distance from the center point to the boundary and the actual temperature at the boundary are recorded. Finally, the thermal field regularity is constructed. The degree to which the heat distribution of the candidate hotspot region conforms to the characteristics of a physical heat source is used to characterize the thermal field regularity. The calculation method is as follows: ; in, The standard deviation of the temperature gradient across all ray directions; As a balance constant, it is set to 1 in this embodiment of the invention; This is a distance correction parameter, which is set to 1 in this embodiment of the invention; Temperature at the center point; This refers to the actual temperature at the boundary. The Euclidean distance from the center point to the boundary; The total number of rays, Indicates the sequence number of the ray.

[0034] To more clearly illustrate thermal field regularity The function and calculation process of this invention will be illustrated by examples in the following embodiments: First, assume that the center temperature of a real overheated region is 80 degrees Celsius, and the boundary temperature is 35 degrees Celsius. The Euclidean distance from the center point to the boundary in each direction is five pixels. At this time, the temperature gradient in each direction is approximately 9. Since the real thermal fault spreads evenly in all directions, the gradients in each direction are very close, so the standard deviation is extremely small. Here, we assume that its standard deviation is 0.2. Substituting it into the formula for thermal field regularity, we can obtain that the thermal field regularity is approximately 50.

[0035] If the interference signal is irregular, the distance from the center point to some directions is very long, resulting in a small temperature gradient, while the distance from the center point to other directions is very short, resulting in a large standard deviation. Here, we assume that the standard deviation is 10. Substituting it into the formula for thermal field regularity, we can obtain that the thermal field regularity is about 5.4, which significantly reduces the value of thermal field regularity.

[0036] Thus, by calculating the thermal field regularity and utilizing the isotropic properties in the physical characteristics, it is possible to effectively assess whether the target area conforms to the diffusion characteristics of a real thermal fault, thereby initially distinguishing between faults and disturbances.

[0037] S3. Perform morphological processing on each candidate hot spot region to determine the body and the outer ring region of the candidate hot spot region, count the total enthalpy within the body and the external overflow heat energy of the outer ring region, and calculate the edge heat capacity transition coefficient used to characterize the degree of energy cutoff at the edge of the region.

[0038] In an optional embodiment, when morphological processing is required for each candidate hotspot region, the candidate hotspot region is first defined as the body and subjected to a morphological dilation operation to generate an outer annular region with a width of a preset number of pixels. Subsequently, an edge thermal capacity transition coefficient is constructed. This is used to distinguish between real thermal faults and reflected interference signals based on differences in physical properties. Edge thermal capacity transition coefficient. The calculation method is as follows: ; in, The ontology representing the candidate hotspot region; The outer ring-shaped area representing the candidate hotspot region; Coordinates within the outer ring area Temperature at that location; The body coordinates of this candidate hotspot region Temperature at that location; The ambient reference temperature; As a positive correction parameter, it is set to 1 in this embodiment of the invention. As the base value, it is set to 5 in this embodiment of the invention; This represents the pixel area of ​​the candidate hotspot region.

[0039] To more clearly illustrate the edge heat capacity transition coefficient The function and calculation process will be illustrated with examples below: In one embodiment of the present invention, it is assumed that the candidate hotspot region being processed is a reflection interference, characterized by: extreme internal heat, and cold air outside the edge of the light spot with a temperature close to the ambient reference temperature. Therefore, the numerator term in the formula for the edge heat capacity transition coefficient is... The cumulative temperature difference is large, assuming it is one thousand, and the denominator term in the formula for the edge heat capacity transition coefficient is... If only a baseline is used, the calculated result of the heat capacity transition coefficient at this edge will be relatively large.

[0040] In another embodiment of the present invention, it is assumed that the candidate hotspot region being processed is a real thermal fault, characterized in that: the peak temperature of the candidate hotspot region is similar to that of the reflected interference, but due to the thermal diffusion effect caused by heat conduction, the outer ring region of the candidate hotspot region is also heated. At this time, the numerator term in the formula for the edge heat capacity transition coefficient is... The cumulative temperature difference is small, and the denominator term in the formula for the edge heat capacity transition coefficient is small. The value of is relatively large, which results in a smaller calculated value for the heat capacity transition coefficient at the edge.

[0041] By comparing the two examples above, it can be seen that the edge thermal capacity transition coefficient is smaller when the candidate hotspot region being processed is a real thermal fault than when the candidate hotspot region being processed is a reflected interference signal.

[0042] Thus, by calculating the edge thermal capacity transition coefficient, and utilizing the thermal diffusion effect of real thermal faults and the energy cutoff characteristics of reflected interference signals, a negative index that is extremely sensitive to reflected interference was constructed, providing a strong basis for subsequent elimination decisions.

[0043] S4. The thermal field regularity and the edge thermal capacity transition coefficient are weighted and fused to calculate the final decision score. When the decision score is greater than the preset decision threshold, the candidate hot spot area is determined to be a real thermal fault; otherwise, it is determined to be an interference signal.

[0044] In an optional embodiment, the present invention constructs a decision score. To determine the nature of candidate hotspot regions, specifically, to calculate decision scores. The method is as follows: ; in, The weighting coefficient for thermal field regularity is set to 0.7 in this embodiment of the invention. The weighting coefficient for the edge heat capacity transition coefficient is set to 0.3 in this embodiment of the invention. This is a bias term.

[0045] To more clearly illustrate the role and calculation process of decision scoring, the embodiments of the present invention will be explained by examples below: Assuming that the candidate hotspot region being processed is a real thermal fault, its thermal field regularity is high, its edge thermal capacity transition coefficient is low, and the final calculated decision score is much greater than 0.5, it is determined to be a real thermal fault. Suppose the candidate hotspot region being processed is a reflected interference signal with low thermal field regularity and a high edge thermal capacity transition coefficient. Since the formula includes a squared term for the edge thermal capacity transition coefficient, a high edge thermal capacity transition coefficient will result in a significant deduction, causing the final decision score to be less than 0.5, thus classifying the candidate hotspot region as interference.

[0046] Thus, by employing nonlinear weighted logic and utilizing the penalty mechanism of the squared term of the edge thermal capacity transition coefficient, the score of interference signals can be rapidly reduced, thereby achieving accurate classification in complex detection environments. Through the fusion and decision-making of multi-dimensional features, the system can physically identify and eliminate false hotspots generated by highly reflective metals, ultimately generating alarm signals and a three-dimensional temperature distribution map, achieving extremely high detection accuracy and anti-interference capabilities.

[0047] Reference Figure 2 The solid line representing a real thermal fault has a bell-shaped waveform that transitions smoothly from the center to the edge, while the dashed line representing metal reflection interference has a plateau-shaped waveform that drops sharply at the edge.

[0048] Reference Figure 3 The region representing a real thermal fault is located in the lower right corner of the figure. This part is characterized by high thermal field regularity but low edge thermal capacity transition coefficient. The region representing a false signal of metal reflection is located in the upper left corner of the figure. This part is characterized by low thermal field regularity but high edge thermal capacity transition coefficient.

[0049] This invention also discloses a fault detection system for distribution box switchgear, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a fault detection method for distribution box switchgear according to the present invention is implemented.

[0050] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0051] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A method for fault detection in distribution box switchgear, characterized in that, include: The original temperature matrix data output by the infrared detector of the distribution box switch cabinet is obtained, the full-image temperature histogram is calculated to determine the ambient reference temperature, and an adaptive floating threshold is set according to the ambient reference temperature. The original temperature matrix data is then binarized and segmented using the adaptive floating threshold, and all connected regions are extracted as candidate hotspot regions. For each candidate hotspot region, polar coordinate sampling is performed. Multiple rays are drawn out from the highest temperature point in the region. The temperature distribution and boundary distance on each ray are retrieved. The thermal field regularity is calculated based on the retrieval results. The thermal field regularity is used to characterize the degree to which the regional heat distribution conforms to the physical heat source characteristics. Morphological processing is performed on each candidate hot spot region to determine the body and the outer ring region of the candidate hot spot region. The total enthalpy within the body and the external overflow heat energy of the outer ring region are statistically analyzed, and the edge heat capacity transition coefficient used to characterize the degree of energy cutoff at the edge of the region is calculated. The thermal field regularity and the edge thermal capacity transition coefficient are weighted and fused to calculate the final decision score. When the decision score is greater than the preset decision threshold, the candidate hot spot area is determined to be a real thermal fault; otherwise, it is determined to be an interference signal.

2. The method for fault detection of distribution box switchgear according to claim 1, characterized in that, The step of setting an adaptive floating threshold based on an environmental reference temperature and using the adaptive floating threshold to perform binarization segmentation on the original temperature matrix data includes: The temperature corresponding to the peak value of the histogram is extracted as the environmental reference temperature. Set a base temperature rise threshold and calculate an adaptive floating threshold, wherein the adaptive floating threshold is the sum of the ambient reference temperature and the base temperature rise threshold; Pixels with temperature values ​​greater than the adaptive floating threshold in the original temperature matrix data are marked as foreground, and pixels with temperature values ​​less than or equal to the adaptive floating threshold are marked as background, thereby extracting the set of connected regions.

3. The method for fault detection of distribution box switchgear according to claim 1, characterized in that, The process of polar coordinate sampling for each candidate hotspot region includes: The highest temperature point within the candidate hotspot area is used as the center point; Construct a polar coordinate system with the center point as the origin, and draw a ray at every preset angle, for a total of multiple rays; Search along each ray direction until the boundary of the candidate hotspot region is encountered, and record the Euclidean distance from the center point to the boundary and the actual temperature at the boundary.

4. The method for fault detection of distribution box switchgear according to claim 3, characterized in that, thermal field regularity Satisfying the relation: ； In the formula, Let the standard deviation of the temperature gradient of all rays be denoted as . It is the equilibrium constant. Temperature at the center point The actual temperature at the boundary. Let Euclidean distance be the distance from the center point to the boundary. For distance correction parameters, The total number of rays, Indicates the sequence number of the ray.

5. The method for fault detection of distribution box switchgear according to claim 1, characterized in that, The morphological processing of each candidate hotspot region includes: Define the candidate hotspot region as the ontology; Perform a morphological dilation operation on the ontology to generate an outer ring region with a preset width.

6. The method for fault detection of distribution box switchgear according to claim 5, characterized in that, The edge heat capacity transition coefficient Satisfying the relation: ； In the formula, Represents the entity, Represents the outer ring area. coordinates within the body The temperature at that location Coordinates within the outer ring area The temperature at that location The ambient reference temperature For positive correction parameters, As the base, The pixel area of ​​the candidate hotspot region.

7. The method for fault detection of distribution box switchgear according to claim 6, characterized in that, The decision score Satisfying the relation: In the formula, and These are the weighting coefficients for thermal field regularity and edge heat capacity transition coefficient, respectively. For thermal field regularity, This is a bias term.

8. The method for fault detection of distribution box switchgear according to claim 7, characterized in that, The weighting coefficient of the thermal field regularity is greater than the weighting coefficient of the edge thermal capacity transition coefficient, and the edge thermal capacity transition coefficient in the formula reflects the nonlinear penalty mechanism through the square term; when the decision score is greater than the preset decision threshold, the target is determined to be a real thermal fault and an alarm signal is generated.

9. A method for fault detection in a distribution box / switch cabinet according to claim 2, characterized in that, The calculation of the full-map temperature histogram to determine the environmental reference temperature includes: Count the frequency of each temperature value in the original temperature matrix data and construct a temperature distribution histogram. The temperature distribution histogram is smoothed to eliminate noise interference; Identify the location of the maximum peak in the smoothed histogram, and determine the corresponding horizontal axis temperature value as the environmental reference temperature to adapt to background temperature changes in different seasons.

10. A fault detection system for distribution box switchgear, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a fault detection method for a distribution box switchgear according to any one of claims 1-9.

Images