Power substation equipment energy efficiency safety detection method and system based on thermal network inversion

Abstract

The present application relates to substation equipment detection technical field, especially in based on the heat network inversion substation equipment energy efficiency safety detection method and system. Through the visible light image and infrared thermal image registration after enhancement fusion processing, make the enhanced fusion image has the temperature precision of infrared image and the texture definition of visible light image, effectively solve the problem that the infrared thermal image texture is fuzzy and the heating component positioning is not accurate; through the heat network model corresponding to the substation equipment, according to the enhanced fusion image, the internal heat source distribution and additional loss value of the substation equipment are inverted, and then the defect category and operation score of the substation equipment are analyzed. Without the experience of operation and maintenance personnel, the accuracy of detection is effectively improved. In addition, the safety defect positioning and energy efficiency loss quantification are output simultaneously. Both have a deep coupling relationship in the physical model, realizing the multidimensional comprehensive evaluation of the equipment state.

Description

Technical Field

[0001] This invention relates to the field of substation equipment testing technology, and in particular to a method and system for energy efficiency and safety testing of substation equipment based on thermal network inversion. Background Technology

[0002] The operational status of substation equipment directly impacts the safe and stable operation of the power grid. Regular inspections of substation equipment, followed by analysis and diagnosis based on their condition, are crucial for ensuring the reliable operation of both equipment and the power grid. Currently, inspection robots collect infrared thermal images of the equipment, which are then used for safety analysis to determine if overheating anomalies exist. However, infrared thermal images only capture the temperature distribution of the equipment's outer surface. Internal defects such as thermal flaws and poor contact cannot be directly detected, requiring maintenance personnel to make indirect inferences based on experience, which introduces uncertainty and the risk of misjudgment. Furthermore, current detection methods are limited in scope, focusing solely on safety status while neglecting energy efficiency assessment. Summary of the Invention

[0003] The main objective of this invention is to provide a method and system for energy efficiency and safety testing of substation equipment based on thermal network inversion. The aim is to be able to invert the distribution of heat sources inside substation equipment, identify the corresponding thermal defect types, and calculate the operating score combined with energy efficiency assessment.

[0004] To achieve the above objectives, the present invention proposes a substation equipment energy efficiency safety testing method based on thermal network inversion, comprising the following steps: Simultaneously acquire visible light and infrared thermal images of substation equipment; Spatial registration of the visible light image and the infrared thermal image yields a registered image pair; The registered image pairs are enhanced and fused to obtain an enhanced fused image; Extract the temperature distribution submatrix of each component in the enhanced fused image; The thermal network model corresponding to the substation equipment is invoked, with the temperature distribution submatrix of each component as input, and the thermal network model outputs the internal heat source distribution and additional loss value of the equipment. Based on the internal heat source distribution of the equipment and the additional loss value, a safety and energy efficiency index analysis is performed to obtain the defect category and operating score of the substation equipment.

[0005] In the above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion, the spatial registration of the visible light image and the infrared thermal image includes the following steps: Feature points in visible light and infrared images are extracted based on the scale-invariant feature transform algorithm, resulting in visible light image feature points and infrared image feature points. Calculate the Euclidean distance between feature points in the visible light image and feature points in the infrared image, and establish feature point pairs using the nearest neighbor matching method; Based on the aforementioned feature point pairs, the homography transformation matrix is ​​estimated using a random sampling consistency algorithm; The infrared image is transformed to the coordinate system of the visible light image according to the homography transformation matrix, and resampling is performed using bilinear interpolation to obtain the registered infrared image.

[0006] In the above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion, the enhanced fusion of the registered image pair includes the following steps: Visible light texture feature map and infrared temperature feature map are extracted from the registered image pair using a feature extraction network; Using guided filtering or conditional generation networks, the infrared temperature feature map is enhanced by using the visible light texture feature map as a condition to obtain an enhanced feature map; The enhanced feature map and the color components of the visible light image are fused to obtain the enhanced fused image.

[0007] In the above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion, extracting the temperature distribution sub-matrix of each component in the enhanced fusion image includes the following steps: The enhanced fused image is segmented into regions using a semantic segmentation network to obtain a classification label map, where each pixel in the classification label map has a corresponding device type identifier and component identifier; Based on the temperature grayscale calibration curve, the grayscale value of each pixel in the enhanced fused image is converted into a temperature value to obtain a temperature distribution matrix; Based on the spatial correspondence between the classification label map and the temperature distribution matrix, the temperature value of the corresponding pixel position of each component is extracted to form the temperature distribution submatrix of that component.

[0008] In the above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion, the thermal network model defines internal heat source nodes, external surface temperature nodes, and thermal resistance and thermal capacity connection relationships between nodes according to the actual structure of the equipment and the actual heat transfer path; the thermal resistance and thermal capacity connection relationships are based on the actual heat transfer path and utilize conduction units, convection units, and radiation units to connect each node. When calling the thermal network model corresponding to the substation equipment, the following steps are included: Mapping component identifiers identified from the enhanced fused image to nodes in the thermal network model; Load the default thermal resistance and thermal capacity parameter set in the thermal network model, and read the current equipment load rate and ambient temperature from the substation monitoring system or field sensors as boundary conditions.

[0009] In the above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion, the output of the internal heat source distribution and additional loss values ​​of the thermal network model includes the following steps: The temperature distribution submatrix is ​​discretized along the surface of the device housing into several local temperature nodes, and the temperature value of each local temperature node is the average value of the pixel temperature in that area; The external surface temperature nodes in the thermal network model are expanded into several local external surface temperature nodes, each corresponding to the local temperature node; the internal heat sources of the equipment are divided into several internal heat source nodes according to their spatial location. Establish a thermal path connection between the internal heat source node and the local external surface temperature node. Based on the internal heat source node, the local external surface temperature node and the corresponding thermal path connection, establish a set of thermal network equations. The set of thermal network equations is to calculate the temperature value of each local external surface temperature node based on the assumed internal heat source node. An inversion objective function is established, which is the sum of squared deviations between the calculated temperature values ​​and the measured temperature values ​​of the local outer surface temperature nodes; The optimal solution for the internal heat source distribution that minimizes the inversion objective function is obtained based on the optimization algorithm. The additional loss value of all internal heat source nodes is obtained by summing the differences between the heat source intensity value of each internal heat source node in the optimal solution of the internal heat source distribution and the corresponding preset heat source intensity value.

[0010] In the above-mentioned substation equipment energy efficiency safety testing method based on heat network inversion, the heat network equations are as follows: ; in: , For node indexing, node For nodes Adjacent nodes, traverse There are 1, 2, 3...M nodes; M=K+N, where M represents the number of nodes, K is the number of internal heat source nodes, and N is the number of local external surface temperature nodes. The internal heat source nodes and the local external surface temperature nodes are uniformly numbered as 1, 2, 3...M. For nodes Temperature; For nodes The heat capacity; For nodes The intensity of the heat source for internal heat source nodes , The intensity of the heat source to be inverted; for local outer surface temperature nodes ,but ; Represents a node With nodes The equivalent thermal resistance between them; Represents nodes The set of directly connected nodes; Represents a node The sum of heat flows transmitted to all adjacent nodes.

[0011] The above-mentioned substation equipment energy efficiency safety testing method based on thermal network inversion includes the following steps for analyzing safety and energy efficiency indicators based on the internal heat source distribution of the equipment and the additional loss value: Extract the regional distribution characteristics of heat sources and the anomaly degree of heat source intensity from the internal heat source distribution; Defect categories are identified based on the distribution characteristics of heat source areas, the anomaly degree of heat source intensity, and the temperature gradient vector to obtain the defect category and the defect severity level, and the corresponding safety score is determined based on the defect severity level; wherein, the temperature gradient vector is obtained by calculating the temperature gradient between the temperature distribution submatrix of the component and the spatial coordinate range of the corresponding component; The energy efficiency degradation rate is obtained by comparing the additional loss value with the rated power of the substation equipment; and the corresponding energy efficiency score is determined based on the energy efficiency degradation rate. The operating score of the substation equipment is obtained by weighting the safety score and the energy efficiency score.

[0012] The above-mentioned substation equipment energy efficiency safety detection method based on thermal network inversion also includes constructing a modular thermal network model library for different equipment types; the modular thermal network model library includes the following three-layer architecture: The base layer includes conduction units, convection units, and radiation units; each unit has standardized input / output interfaces; for example, the input is temperature difference / heat flux, and the output is heat flux / temperature difference; wherein, the conduction unit is: ; ; in, For conductive thermal resistance, This is the length of the heat transfer path. The thermal conductivity of the material. This refers to the heat transfer cross-sectional area; For heat capacity, For material density, For specific heat capacity, Unit volume; The convection unit is: ; in, For convective thermal resistance, The convective heat transfer coefficient is... For heat exchange area; The radiating element is: ； in The Stefan-Boltzmann constant is... For surface emissivity, For the radiation area, , These are the absolute temperatures of the radiating surface and the receiving surface, respectively. The equipment layer constructs a corresponding modular thermal network model based on the structure and heat transfer path of different types of equipment. The parameter layer is configured with a default set of thermal resistance and thermal capacity parameters for each modular thermal network model, and has an online calibration interface for model parameters; it is used to read the current equipment load rate and ambient temperature from the substation monitoring system or field sensors as boundary conditions.

[0013] Before calling the thermal network model corresponding to the substation equipment, the corresponding thermal network model is selected from the modular thermal network model library based on the equipment identifier identified from the enhanced fusion image.

[0014] The second aspect of this application discloses a substation equipment energy efficiency and safety testing system based on thermal network inversion, applied to the aforementioned substation equipment energy efficiency and safety testing method based on thermal network inversion, including: The acquisition module is used to simultaneously acquire visible light and infrared thermal images of substation equipment; The registration module is used to spatially register the visible light image and the infrared thermal image to obtain a registered image pair; An enhancement module is used to enhance and fuse the registered image pairs to obtain an enhanced fused image; The extraction module is used to extract the temperature distribution sub-matrix of each component in the enhanced fused image; The inversion module is used to call the thermal network model corresponding to the substation equipment. The thermal network model takes the temperature distribution submatrix of each component as input and outputs the internal heat source distribution and additional loss value of the equipment. The analysis module is used to perform safety and energy efficiency index analysis based on the internal heat source distribution of the equipment and the additional loss value, so as to obtain the defect category and operating score of the substation equipment.

[0015] The technical solution provided by this invention may include the following beneficial effects: By registering visible light images and infrared thermal images and then performing enhancement and fusion processing, the enhanced and fused image combines the temperature accuracy of infrared images with the texture clarity of visible light images, effectively solving the problem of inaccurate positioning of heating components caused by blurred texture in infrared thermal images. By using a thermal network model corresponding to the substation equipment, the internal heat source distribution and additional loss values ​​of the substation equipment are derived from the enhanced fusion image, thereby analyzing the defect categories and operational scores of the substation equipment. This eliminates the need for maintenance personnel to make indirect inferences based on experience, effectively improving the accuracy of detection. Furthermore, it simultaneously outputs the location of safety defects and the quantification of energy efficiency losses. Both are deeply coupled within the physical model (both are results caused by heat source distribution), enabling a multi-dimensional comprehensive evaluation of the equipment's condition. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the energy efficiency and safety testing method for substation equipment based on thermal network inversion according to the present invention. Figure 2 This is a schematic diagram of the process for spatially registering visible light images and infrared thermal images in the energy efficiency and safety testing method of the present invention; Figure 3 This is a schematic diagram of the process for enhancing the fusion and registration of image pairs in the energy efficiency and safety testing method of the present invention; Figure 4 This is a schematic diagram of the process for extracting the temperature distribution sub-matrix in the energy efficiency and safety testing method of the present invention; Figure 5 This is a schematic diagram of the process for inverting the distribution of internal heat sources in the energy efficiency and safety testing method of the present invention; Figure 6 This is a schematic diagram of the framework of the substation equipment energy efficiency safety detection system based on thermal network inversion according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0020] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0021] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the word "and / or" throughout the text means including three parallel solutions; taking "A and / or B" as an example, it includes solution A, solution B, or a solution that simultaneously satisfies A and B. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0022] The following is combined with Figure 1 The present invention describes a method for energy efficiency and safety testing of substation equipment based on thermal network inversion, comprising the following steps: Step S1: Simultaneously acquire visible light and infrared thermal images of the substation equipment; the substation equipment may include main transformers, GIS circuit breakers, disconnectors, instrument transformers, or surge arresters, etc. For example, a patrol robot equipped with a dual-light camera can acquire visible light and infrared thermal images of the substation equipment. Further, taking a main transformer as an example, the patrol robot acquires images of the outer wall of the main transformer tank and the bushing roots. If the substation equipment is a GIS circuit breaker, the patrol robot acquires images near the observation window of the circuit breaker's gas chamber. If the equipment is a disconnector, images of the moving and stationary contact areas of the disconnector can be acquired.

[0023] Step S2: Spatially register the visible light image and the infrared thermal image to obtain a registered image pair; thus, the two images are transformed to a unified spatial coordinate system. It is worth noting that since the visible light image and the infrared thermal image are acquired synchronously, they are essentially aligned in terms of time reference. That is, in the registered image pair, the visible light image and the infrared thermal image are unified in both time and space.

[0024] Step S3: Enhance and fuse the registered image pairs to obtain an enhanced fused image. Through enhancement and fusion processing, the processed image combines the temperature accuracy of an infrared image with the texture clarity of a visible light image.

[0025] Step S4: Extract the temperature distribution sub-matrix of each component in the enhanced fused image; specifically, in the enhanced fused image, the gray value of each pixel can be converted into a temperature value, and the temperature distribution sub-matrix of the component can be constructed based on the temperature value corresponding to the component and space.

[0026] Step S5: Invoke the thermal network model corresponding to the substation equipment, using the temperature distribution sub-matrix of each component as input. The thermal network model outputs the internal heat source distribution and additional loss values ​​of the equipment. It should be understood that a thermal network model is an engineering method used to simulate and analyze heat transfer behavior in complex systems. It simplifies the heat conduction, convection, and radiation processes of a physical structure into a network composed of elements such as thermal resistance and thermal capacity. In this application, the temperature distribution sub-matrix of each component is used as input to the thermal network model, and the thermal network model outputs the internal heat source distribution and additional loss values ​​of the equipment.

[0027] Step S6: Based on the internal heat source distribution of the equipment and the additional loss value, perform safety and energy efficiency index analysis to obtain the defect category and operating score of the substation equipment, so as to achieve safety detection.

[0028] This invention enhances and fuses visible light and infrared thermal images after registration, resulting in an image that combines the temperature accuracy of infrared images with the texture clarity of visible light images. This effectively solves the problem of inaccurate positioning of heat-generating components caused by blurred textures in infrared thermal images. By using a thermal network model corresponding to the substation equipment, the internal heat source distribution and additional loss values ​​of the substation equipment are derived from the enhanced and fused image, thereby analyzing the defect categories and operational scores of the substation equipment. This eliminates the need for maintenance personnel to make indirect inferences based on experience, effectively improving detection accuracy. Furthermore, it simultaneously outputs the location of safety defects and the quantification of energy efficiency losses. Both are deeply coupled within the physical model (both are results caused by heat source distribution), enabling a multi-dimensional comprehensive evaluation of equipment status.

[0029] In one specific embodiment, the spatial registration of the visible light image and the infrared thermal image includes the following steps: Step S21: Extract visible light images based on the Scale Invariant Feature Transform (SIFT) algorithm and infrared images From the feature points in the image, we obtain the feature points of the visible light image and the feature points of the infrared image; Step S22: Calculate the Euclidean distance between the feature points in the visible light image and the feature points in the infrared image, and establish feature point pairs using the nearest neighbor matching method; Step S23: Based on the feature point pairs, estimate the homography transformation matrix using the Random Sample Consensus (RANSAC) algorithm; Step S24: Transform the infrared image to the coordinate system of the visible light image according to the homography transformation matrix, and resample using bilinear interpolation to obtain the registered infrared image. That is, to obtain the registered image pair. .

[0030] In one specific embodiment, the enhancement and fusion of the registered image pair includes the following steps: Step S31: Extract visible light texture feature maps from the registered image pair using a feature extraction network. Infrared temperature feature map Specifically, feature extraction can be achieved based on convolutional neural networks. For example, visible light texture feature maps... Infrared temperature feature maps can be extracted by combining the Sobel operator with deep convolutional layers, including spatial structure information such as device edges and surface textures. The temperature gradient distribution on the device surface can be extracted through the temperature grayscale mapping relationship of infrared images.

[0031] Step S32: Using guided filtering or a conditional generation network, the infrared temperature feature map is enhanced based on the visible light texture feature map to obtain an enhanced feature map; specifically, taking guided filtering as an example, the visible light texture feature map is... As a condition, infrared temperature feature map Guided filtering is applied to enhance the feature map so that it retains the infrared temperature attributes while inheriting the edge and texture details of the visible light image. More specifically, the mathematical expression of guided filtering is: ; This represents the enhanced feature map. Visible light texture feature map The linear coefficients obtained to guide the calculation.

[0032] For example, taking a conditional generative network as an example, a generative network with an encoder-decoder structure is constructed to generate infrared temperature feature maps. Input encoder to visible light texture feature map As conditional inputs, the features are fused through conditional normalization or feature concatenation, and the decoder outputs an enhanced feature map. .

[0033] Step S33: Fuse the enhanced feature maps The color components (chroma, saturation) of the visible light image are used to generate an enhanced fused image through an image reconstruction network. Thus, by "actively guiding" infrared image enhancement using visible light texture features, the problem of blurred texture in infrared images is fundamentally solved. This active guidance mechanism enables the enhanced image to retain infrared temperature accuracy while inheriting edge and texture information from visible light, providing high-quality boundary conditions for subsequent inversion.

[0034] In a specific embodiment, extracting the temperature distribution sub-matrix of each component in the enhanced fused image includes the following steps: Step S41: The enhanced fused image is segmented into regions using a semantic segmentation network to obtain a classification label map. Each pixel in the classification label map has a corresponding device type identifier and component identifier. For example, a semantic segmentation network, such as DeepLabV3+ or Mask R-CNN, is used to enhance the fused image. As input, enhance the fused image The system classifies each pixel individually, identifies and distinguishes device types and key components, and obtains a pixel-level classification label map. For example, if the equipment type is any one of transformer, circuit breaker, GIS, disconnector, instrument transformer, or surge arrester, the key components of a transformer include: tank wall, heat sink, bushing top, and tap changer. The key components of a disconnector include: air chamber, flange interface, and observation window. The key components of a disconnector include: contacts, contact fingers, and lead connectors. Further explanation is provided in the classification label diagram. Each pixel in the image is marked with a corresponding device type identifier and component identifier, such as transformer-tank wall, transformer-heat sink, etc.

[0035] Step S42: Based on the temperature grayscale calibration curve of the infrared thermal image, convert the grayscale value of each pixel in the enhanced fused image into a temperature value to obtain a temperature distribution matrix. The temperature-grayscale calibration curve is a grayscale value-temperature mapping curve established in advance using the factory calibration parameters of the infrared thermal imager or on-site blackbody calibration. Specifically, for each infrared camera, before the equipment is put into operation, calibration is performed using a blackbody radiation source with a known temperature (e.g., a temperature range of 30°C to 150°C, with data collected at 10°C intervals), and the functional relationship between grayscale value and temperature is fitted to obtain the result. Typically, polynomial fitting (order 2 to 3) or piecewise linear interpolation can be used. Therefore, based on the temperature grayscale calibration curve, the grayscale value of each pixel in the enhanced fused image can be converted into a temperature value, resulting in a refined temperature distribution matrix. .

[0036] Step S43: Based on the spatial correspondence between the classification label map and the temperature distribution matrix, extract the temperature value of the corresponding pixel position for each component to form a temperature distribution sub-matrix for that component. .

[0037] In one specific embodiment, the thermal network model is defined based on the actual structure of the device and the actual heat transfer path. Internal heat source nodes, The external surface temperature nodes and the thermal resistance and thermal capacity connection relationships between the nodes; the thermal resistance and thermal capacity connection relationships are based on the actual heat transfer path and utilize conduction units, convection units and radiation units to connect each node; When calling the thermal network model corresponding to the substation equipment, the following steps are included: Step S51: Map the component identifiers identified from the enhanced fusion image to the nodes in the thermal network model; for example, the “core” node in the thermal network model of a transformer corresponds to the actual identified component “core”.

[0038] Step S52: Load the default thermal resistance and thermal capacity parameter set in the thermal network model, and read the current equipment load rate from the substation monitoring system or field sensors. (e.g., transformer active power, circuit breaker current) and ambient temperature As boundary conditions, the thermal resistance and thermal capacity parameter set is calculated based on the equipment's rated parameters, material properties, and structural dimensions. Specific methods for obtaining these parameters include: Nameplate parameter conversion: Based on the rated capacity, voltage level, cooling method, etc. on the equipment nameplate, and combined with the empirical formulas in standard design manuals (such as "Power Transformer Design Manual" and "High Voltage Switchgear Design Code"), estimate the initial values ​​of thermal resistance and thermal capacity of each node. Factory test data: Use the temperature rise test report when the equipment leaves the factory to deduce key thermal resistance parameters; Historical operational data statistics: For equipment in operation, historical infrared temperature measurement data and load records are used to calibrate the default parameters through parameter identification methods (such as least squares fitting).

[0039] For example, taking the thermal network model of a transformer as an example, a five-node model of "core-winding-insulating oil-tank-shell" is adopted. In this model, the internal heat source nodes are the core and the winding. =2; In the thermal network model, the outer surface temperature nodes are the tank wall and outer shell (which can be further subdivided into heat sink areas), and the outer surface temperature nodes... =2~4. The thermal resistance and thermal capacity connection relationships between nodes include: the connection between the core and the winding is through a conduction unit, and the thermal resistance is determined by the contact area and the thermal conductivity of the contact material; the connection between the winding and the insulating oil is through a convection unit, and the convective heat transfer coefficient depends on the oil flow velocity (which can be estimated based on the load rate); the connection between the insulating oil and the tank wall is through a convection unit, and the inner wall of the tank exchanges heat with the oil through convection; the connection between the tank wall and the outer shell (heat sink) is through a conduction unit, and the tank wall is made of metal with high thermal conductivity, so the thermal resistance can be ignored or simplified; both convection units (natural / forced convection of external air) and radiation units (radiation from the outer shell) are set between the outer shell and the external environment. The thermal capacity of each node is calculated based on its material volume and specific heat capacity, which can be calculated according to a preset thermal capacity formula. The initial values ​​of thermal resistance and thermal capacity are preset according to the rated capacity of the equipment and nameplate parameters (such as oil volume and core weight).

[0040] In one specific embodiment, the output of the thermal network model of the internal heat source distribution and additional loss values ​​of the device includes the following steps: Step S53: Discretize the temperature distribution sub-matrix along the surface of the device casing into several local temperature nodes. The temperature value of each local temperature node is the average value of the pixel temperature in that area, denoted as . The subscripts p=1,2,...,N represent the index numbers of the local temperature nodes; for example, if it is the tank wall of a transformer, it can be divided along the height direction. Layers, divided circumferentially Number of sectors, total Each node; for a GIS air chamber, it is divided along the axial direction. Segment, divided along the circumference Each sector.

[0041] Step S54: Expand the outer surface temperature node in the thermal network model into several local outer surface temperature nodes, each of which corresponds to the local temperature node; that is, the number of local outer surface temperature nodes is the same as the number of local temperature nodes, both being N.

[0042] The internal heat sources of the equipment are divided according to their spatial location. Internal heat source nodes The number and initial definition of internal heat source nodes are derived from the predefined internal heat source nodes in the modular thermal network model (such as the "core" and "winding" nodes of a transformer, and the "contacts" and "arc extinguishing chamber" nodes of a circuit breaker). Of course, this step can be further refined according to the actual structure. For example, for transformer windings, they can be divided into multiple sub-nodes along the height direction (such as upper winding, middle winding, and lower winding); for multiple sets of contact fingers of a disconnecting switch, each set of contact fingers is an independent heat source node.

[0043] Step S55: Establish thermal path connections between internal heat source nodes and the local external surface temperature nodes. Each local external surface temperature node establishes a direct thermal path connection with the nearest spatially located internal heat source node. The spatial correspondence is predetermined by the structural geometry of the equipment and the heat transfer path. If the influence range of a certain internal heat source node covers multiple local external surface temperature nodes, then the internal heat source node establishes thermal path connections with each of the multiple local external surface temperature nodes, forming a "one-to-many" radial topology. A set of thermal network equations is established based on the internal heat source nodes, local external surface temperature nodes, and corresponding thermal path connections. The set of thermal network equations calculates the temperature values ​​of each local external surface temperature node using the assumed internal heat source node.

[0044] Step S56: Establish the inversion objective function The inversion objective function This is the sum of squares of the deviations between the calculated temperature value and the measured temperature value of the local outer surface temperature node; the specific formula is: ; in, This indicates that the heat source intensity of each internal heat source node is assumed to be... At that time, the calculated temperature value of the p-th local outer surface temperature node obtained from the thermal network equations; This represents the measured temperature value at a local temperature node.

[0045] Step S57: Solve for the optimal solution of the internal heat source distribution that minimizes the inversion objective function based on an optimization algorithm; specifically, the gradient descent method or the example group optimization algorithm can be used to solve for the optimal solution of the internal heat source distribution that minimizes the inversion objective function. For example, using the current equipment load rate... As input. For each internal heat source node, its heat source intensity value under normal conditions. With current equipment load rate The relationship can be pre-fitted using equipment factory test data or historical normal operation data. Therefore, when setting the initial values, the initial heat source intensity value of each internal heat source node is taken as... This refers to the heat source intensity of the internal heat source node under normal conditions at the current load rate of the equipment. If historical data is unavailable, the normal heat source intensity at the rated load rate can be multiplied by [the value of the heat source intensity]. Estimate (resistive losses).

[0046] Calculate the objective function Heat source intensity value of each internal heat source node The gradient is used to update the heat source intensity values ​​of each internal heat source node along the gradient descent direction until the objective function is reached. Convergence or reaching the maximum number of iterations.

[0047] Step S58: Sum the difference between the heat source intensity value of each internal heat source node in the optimal solution of the internal heat source distribution and the corresponding preset heat source intensity value to obtain the additional loss value of all internal heat source nodes.

[0048] For example, the preset heat source intensity value is the initial value of the internal heat source node, i.e. The formula for calculating the additional loss value is: ; in, Let be the heat source intensity value of the i-th internal heat source node in the optimal solution of the internal heat source distribution; This is the initial value for the internal heat source node.

[0049] This invention extends the traditional thermal network model from "point temperature" modeling to "field temperature" modeling by using boundary condition decomposition, local node coupling, and optimization solution techniques, achieving deep coupling between the refined temperature distribution in the fusion image and the physical model of the thermal network.

[0050] More specifically, the heat network equations are: ; in: , For node indexing, node For nodes Adjacent nodes, traverse There are 1, 2, 3...M nodes; M=K+N, where M represents the number of nodes, K is the number of internal heat source nodes, and N is the number of local external surface temperature nodes. The internal heat source nodes and the local external surface temperature nodes are uniformly numbered as 1, 2, 3...M. For nodes Temperature; For nodes The heat capacity; For nodes The intensity of the heat source for internal heat source nodes , The intensity of the heat source to be inverted; for local outer surface temperature nodes ,but ; Represents a node With nodes The equivalent thermal resistance between them; Represents nodes The set of directly connected nodes; Represents a node The sum of heat flows transmitted to all adjacent nodes.

[0051] In a specific embodiment, the safety and energy efficiency index analysis based on the internal heat source distribution of the device and the additional loss value includes the following steps: Step S61: Extract the heat source region distribution characteristics and heat source intensity anomaly degree from the internal heat source distribution; for example, the calculation formula for the heat source intensity anomaly degree of each internal heat source node is: .like If so, the heat source node is determined to be abnormal. To set the source anomaly threshold, a normal value is taken. The 20% threshold is determined based on statistical analysis of 200 sets of historical fault data. When the anomaly rate exceeds 20%, the probability of equipment defects significantly increases to over 85%. Internal heat source nodes are bound to the spatial coordinates within the equipment during local node coupling. Therefore, the distribution characteristics of heat source areas include equipment identification, component identification, and the corresponding regional spatial coordinates. For example, "Upper part of the low-voltage winding of phase B of main transformer No. 1 (x=1.2m, y=2.5m, z=0.8m)".

[0052] Step S62: Based on the distribution characteristics of the heat source region, the anomaly degree of the heat source intensity, and the temperature gradient vector Defect category identification is performed to obtain the defect category and defect severity level, and the corresponding safety score is determined based on the defect severity level; wherein, the temperature gradient vector The temperature gradient is calculated by using the temperature distribution submatrix of the component and the spatial coordinate range of the corresponding component. For example, defect types include point heat source defects, area heat source defects, cluster heat source defects, and interphase difference defects. Further, for example, the defect type can be comprehensively determined based on the table below.

[0053] The severity level of a defect includes attention level, abnormal level, severe level, and crisis level, which can be judged based on the following conditions.

[0054] When the heat source intensity anomaly is: The system is classified as Level 90, meaning it can continue operating but requires enhanced monitoring; the corresponding safety score is 90. When the heat source intensity anomaly is: It is classified as an abnormal level, indicating a significant defect, and is scheduled for maintenance soon; the corresponding safety score is 70.

[0055] When the heat source intensity anomaly is: The assessment is severe, indicating a serious defect; repairs should be arranged as soon as possible. The corresponding safety score is 50.

[0056] When the heat source intensity anomaly is: The situation is assessed as crisis level, indicating a risk of an accident. Power should be cut off immediately for maintenance; the corresponding safety score is 30.

[0057] Step S63: Obtain the energy efficiency degradation rate by comparing the additional loss value with the rated power of the substation equipment; and determine the corresponding energy efficiency score based on the energy efficiency degradation rate; optionally, the formula for calculating the energy efficiency degradation rate is: ;in Rated power of the equipment (corresponding to rated load rate) (Rated loss below). Note that... Current load rate The additional losses are approximately proportional to the square of the load factor (for resistive losses), therefore, actual energy efficiency assessments should be normalized to incorporate the load factor. In an optional embodiment, 100 minutes, At 0:00, linear interpolation is used in the middle. Of course, the annual additional electricity consumption can also be calculated based on the additional loss value; the calculation formula is: in, This indicates the number of hours the equipment operates per year.

[0058] Step S64: Calculate the substation equipment's operating score by weighting the safety score and energy efficiency score. Specifically, the calculation formula is as follows: ; in, Indicates the safety status score. Indicates energy efficiency rating. Indicates the weighting coefficient. .

[0059] In other embodiments, the method further includes building a modular thermal network model library for different device types; the modular thermal network model library includes the following three-layer architecture: The base layer includes conduction units, convection units, and radiation units; each unit has standardized input / output interfaces; for example, the input is temperature difference / heat flux, and the output is heat flux / temperature difference; wherein, the conduction unit is: ; ; in, For conductive thermal resistance, This is the length of the heat transfer path. The thermal conductivity of the material. The heat transfer cross-sectional area (m²) is the heat transfer cross-sectional area. For heat capacity, For material density, For specific heat capacity, Unit volume; The convection unit is: ; in, For convective thermal resistance, The convective heat transfer coefficient is (W / (m²·K)). The heat exchange area is (m²). The radiating element is: ; in The Stefan-Boltzmann constant is... The value is ), For surface emissivity, The value range of is (0 to 1). The radiation area (m²) , Here, K represents the absolute temperatures (K) of the radiating and receiving surfaces, respectively; when the temperature variation range is small, the radiating element can be approximated as linearized as... ,in This represents the average temperature.

[0060] The equipment layer constructs a corresponding modular thermal network model based on the structure and heat transfer path of different types of equipment. For example, in the modular transformer thermal network model, a five-node model of "core-winding-insulating oil-tank-shell" is adopted.

[0061] Node definitions: ① Iron core, ② Winding (which can be further subdivided into high-voltage winding and low-voltage winding), ③ Insulating oil (oil top and oil bottom can be layered), ④ Tank wall, ⑤ Shell (heat sink area is modeled separately).

[0062] Thermal circuit connections: Between the iron core and the winding: connected through a conduction unit, the thermal resistance is determined by the contact area and the thermal conductivity of the contact material; Between the winding and the insulating oil: connected through a convection unit, the convective heat transfer coefficient depends on the oil flow velocity; Between the insulating oil and the tank wall: connected through a convection unit, the inner wall of the tank exchanges heat with the oil through convection; Between the tank wall and the outer shell (heat sink): connected through a conduction unit, the tank wall is made of metal with high thermal conductivity, the thermal resistance can be ignored or simplified; Between the outer shell and the external environment: both convection units (natural / forced convection of external air) and radiation units (radiation from the outer shell) are set up.

[0063] Heat capacity allocation: The heat capacity of each node is calculated based on its material volume and specific heat capacity, corresponding to the heat capacity formula of the base layer.

[0064] Parameter initialization: Preset the initial values ​​of thermal resistance and thermal capacity based on the equipment's rated capacity and nameplate parameters (such as oil volume and core weight).

[0065] For example, in the modular GIS circuit breaker thermal network model, a four-node model of "contacts - arc-extinguishing chamber - SF6 gas - casing" is adopted; Node definitions: ① Contact (moving and stationary contacts can be modeled separately or merged into an internal heat source node), ② Arc extinguishing chamber (including arc contacts and surrounding structures), ③ SF6 gas (gas inside the chamber, which can be regarded as a lumped node), ④ Outer shell (outer wall of the chamber).

[0066] Thermal connections: Between the contacts and the arc-extinguishing chamber: connected via a conduction unit, the thermal resistance being determined by the contact thermal resistance between the contacts and the arc-extinguishing chamber structure; Between the arc-extinguishing chamber and SF6 gas: connected via a convection unit, the convective heat transfer coefficient of SF6 gas depends on the gas pressure (typically 0.5–0.7 MPa) and flow state; Between SF6 gas and the outer shell: connected via a convection unit (convective heat transfer between the gas and the inner wall of the outer shell); Between the outer shell and the external environment: both a convection unit (for external air) and a radiation unit are provided.

[0067] Parameter initialization: Preset parameters based on circuit breaker rated current, SF6 gas pressure, casing material, etc.

[0068] For example, in the modular disconnector thermal network model, a four-node model of "contact - contact finger - lead wire - shell" is adopted, and multiple sets of contact fingers are discretized into multiple parallel heat flow channels.

[0069] Node definition: ① Contact (stationary contact), ② Contact finger group (discrete according to the actual number of contact fingers) Each independent node Take 3 to 8), ③ lead wire (terminal and lead wire), ④ housing (equipment housing).

[0070] Thermal circuit connections: Between the contact and each set of contact fingers: connected via conductive units; contact thermal resistance depends on clamping force (related to spring pressure); contact thermal resistance is calculated independently for each set of contact fingers. Between each set of contact fingers and the lead wire: connected via conductive units; contact thermal resistance between the contact fingers and the lead wire. Between the lead wire and the housing: connected via conductive units (thermal resistance of the insulating support). Between the housing and the external environment: both convection and radiation units are used. Parallel channel processing: The heat flow paths of multiple sets of contact fingers are represented as a parallel structure in the thermal network model. Each set of contact fingers forms an independent branch with the contact and lead wire; the thermal resistances of each branch are in parallel, and the total heat flow is the sum of the heat flows of each branch. The introduction of parallel channels allows for the independent solution of the contact state of each set of contact fingers during inversion, thereby accurately locating the clamping force anomaly of a single contact finger.

[0071] Parameter initialization: The initial contact thermal resistance value is preset based on the rated current of the disconnector switch, the contact index, the spring design parameters, etc.

[0072] The parameter layer is configured with a default set of thermal resistance and thermal capacity parameters for each modular thermal network model, and has an online calibration interface for model parameters; it is used to read the current equipment load rate and ambient temperature from the substation monitoring system or field sensors as boundary conditions.

[0073] Before calling the thermal network model corresponding to the substation equipment, the corresponding thermal network model is selected from the modular thermal network model library based on the equipment identifier identified from the enhanced fusion image.

[0074] In this way, a layered and modular thermal network model library of "foundation layer - equipment layer - parameter layer" was constructed, solving the problem of the large variety of substation equipment and the difficulty in unified management and calling of models. This architecture enables the system to automatically call the corresponding modular thermal network model based on the equipment type identified by semantic segmentation, without manual intervention, and provides support for the automation and intelligence of collaborative detection.

[0075] For example, in one specific embodiment of the present invention, the detection of the contact status of a GIS circuit breaker is taken as an example.

[0076] Step S1: At a 500kV GIS substation, an inspection robot is used to collect images near the observation window of the circuit breaker gas chamber, obtaining visible light images. Infrared thermal image .

[0077] Step S2: Image registration: Input: Visible light image Infrared thermal image ; The processing steps include: SIFT feature point extraction and RANSAC transform matrix estimation for registration, with the reprojection error threshold set to 3 pixels; Output: Registered visible light image and infrared images Pixel-level alignment.

[0078] Step S3: Image Enhancement and Fusion Input: Registered image pairs ; The processing steps include: employing a conditional generative adversarial network (cGAN) with edge maps extracted from visible light images as conditions; and an encoder extracting infrared temperature features. The decoder uses visible light texture features (Edge map) is used as a condition for feature fusion; the loss function includes a temperature fidelity term and a texture consistency term to ensure that the enhanced image retains infrared temperature accuracy and visible light texture details; Output: Enhanced fused image The structure, such as the flange joint of the air chamber and the edge of the observation window, is clearly visible, and the temperature distribution corresponds precisely to the outer shell structure.

[0079] Step S4: Device identification and refined temperature extraction: Input: Enhanced fused image .

[0080] Processing: A semantic segmentation network (DeepLabV3+) performs pixel-by-pixel classification, identifying it as a "GIS circuit breaker," and further segments the outer casing area corresponding to the three-phase gas chamber; based on the infrared calibration curve (factory calibration polynomial)... (On-site blackbody verification and correction) converts grayscale values ​​to temperature values; extracts the temperature distribution matrix of each phase gas chamber shell. The temperature in the central region of phase A is significantly higher (72°C), while phases B and C are approximately 55°C and uniformly distributed. The temperature gradient vector was calculated. The temperature gradient of phase A points towards the internal contact position, with a gradient magnitude of approximately 5.6°C / m.

[0081] Output: Equipment type "GIS circuit breaker"; Component identification “A phase gas chamber”, “B phase gas chamber”, “C phase gas chamber”; Temperature distribution matrix of each component , , ; Temperature gradient eigenvector , , .

[0082] Step S5: Model Invocation: Input: Device type "GIS circuit breaker", current load current (Rated current) Load rate Ambient temperature (Read from SCADA system); The processing includes: the system automatically calls the "GIS circuit breaker modular thermal network model" (four nodes: contacts, arc-extinguishing chamber, SF6 gas, and casing). The contacts and arc-extinguishing chamber are the internal heat source nodes. The outer casing is the outer surface temperature node (extended in step S6). Default thermal resistance and thermal capacity parameters are loaded (based on contact resistance at rated current, SF6 gas pressure of 0.6 MPa, casing material, etc.). Output: Modular thermal network model Includes internal heat source nodes (Main contact) (Arc contact / arc extinguishing chamber), external surface temperature nodes, thermal resistance and thermal capacity network connections, load factor Ambient temperature .

[0083] Distributed boundary condition inversion: Input: Temperature distribution matrix , , Hot network model Load rate Ambient temperature .

[0084] The processing includes: boundary condition decomposition: discretizing the temperature distribution of each phase gas chamber shell along the axial direction. The segment, circumferentially discretized as Number of sectors, total Local temperature nodes The original single outer shell temperature node is expanded to 96 local outer shell nodes, each of which establishes a thermal path connection with the spatially nearest internal heat source node. The gradient descent method is used to invert the internal heat source distribution, with the objective function... The initial value is set to (Under normal conditions, the heat source value is proportional to the square of the load rate).

[0085] Output: Internal heat source distribution (A-phase main contact) B-phase main contact C-phase main contact The contact resistance of the main contact of phase A is obtained by inversion. (Normal value) Equivalent heat source anomaly The contact resistance between phases B and C is normal (0.52–0.58 mΩ). Additional losses... .

[0086] Step S6: Collaborative Output: Input: Internal heat source distribution Additional losses ; Safety status assessment: Location of abnormal heat source: Phase A main contacts Anomaly ; Defect type determination: Heat source area: point-like (contact point area < 10% of total contact area); Heat source intensity: Anomaly degree 145%, significantly higher than adjacent nodes (B, %); Temperature gradient: The gradient magnitude is 5.6°C / m, and the direction is towards the center of the contact, which is consistent with the characteristics of "point heat source (poor contact)". Severity level classification (based on step 6.1.3): It was classified as "serious"; Energy efficiency index calculation: Efficiency degradation rate: ; Annual additional power loss: The circuit breaker is calculated based on actual load duration, with approximately 2000 operation times per year, equivalent to approximately 2000 hours of full-load operation. ; Overall Health Index: Safety rating: Severity level corresponds to 50 points; Energy efficiency rating: 100 points; point; Output: Safety Status Report: A phase has a "poor contact of main contacts" defect, located at "GIS5011 circuit breaker A phase main contacts". , , The severity level is "Severe". Energy efficiency report: Efficiency degradation rate 0.0030%, annual additional electricity loss 1,480 kWh / year.

[0087] The present invention also provides a substation equipment energy efficiency and safety testing system 100 based on thermal network inversion, applied to the substation equipment energy efficiency and safety testing method based on thermal network inversion described in any of the above examples, comprising: Acquisition module 101 is used to simultaneously acquire visible light images and infrared thermal images of substation equipment; Registration module 102 is used for spatiotemporal registration of the visible light image and the infrared thermal image to obtain a registered image pair; Enhancement module 103 is used to enhance and fuse the registered image pair to obtain an enhanced fused image; Extraction module 104 is used to extract the temperature distribution sub-matrix of each component in the enhanced fused image; The inversion module 105 is used to call the thermal network model corresponding to the substation equipment, taking the temperature distribution submatrix of each component as input, and the thermal network model outputs the internal heat source distribution and additional loss value of the equipment. Analysis module 106 is used to perform safety and energy efficiency index analysis based on the internal heat source distribution of the equipment and the additional loss value, and to obtain the defect category and operating score of the substation equipment.

[0088] This invention enhances and fuses visible light and infrared thermal images after registration, resulting in an image that combines the temperature accuracy of infrared images with the texture clarity of visible light images. This effectively solves the problem of inaccurate positioning of heat-generating components caused by blurred textures in infrared thermal images. By using a thermal network model corresponding to the substation equipment, the internal heat source distribution and additional loss values ​​of the substation equipment are derived from the enhanced and fused image, thereby analyzing the defect categories and operational scores of the substation equipment. This eliminates the need for maintenance personnel to make indirect inferences based on experience, effectively improving detection accuracy. Furthermore, it simultaneously outputs the location of safety defects and the quantification of energy efficiency losses. Both are deeply coupled within the physical model (both are results caused by heat source distribution), enabling a multi-dimensional comprehensive evaluation of equipment status.

[0089] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made under the concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A method for energy efficiency and safety testing of substation equipment based on thermal network inversion, characterized in that: Includes the following steps: Simultaneously acquire visible light and infrared thermal images of substation equipment; Spatial registration of the visible light image and the infrared thermal image yields a registered image pair; The registered image pairs are enhanced and fused to obtain an enhanced fused image; Extract the temperature distribution submatrix of each component in the enhanced fused image; The thermal network model corresponding to the substation equipment is invoked, with the temperature distribution submatrix of each component as input, and the thermal network model outputs the internal heat source distribution and additional loss value of the equipment. Based on the internal heat source distribution of the equipment and the additional loss value, a safety and energy efficiency index analysis is performed to obtain the defect category and operating score of the substation equipment.

2. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: The spatially registered visible light image and the infrared thermal image include the following steps: Feature points in visible light and infrared images are extracted based on the scale-invariant feature transform algorithm, resulting in visible light image feature points and infrared image feature points. Calculate the Euclidean distance between feature points in the visible light image and feature points in the infrared image, and establish feature point pairs using the nearest neighbor matching method; Based on the aforementioned feature point pairs, the homography transformation matrix is ​​estimated using a random sampling consistency algorithm; The infrared image is transformed to the coordinate system of the visible light image according to the homography transformation matrix, and resampling is performed using bilinear interpolation to obtain the registered infrared image.

3. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: The enhanced fusion of the registered image pair includes the following steps: Visible light texture feature map and infrared temperature feature map are extracted from the registered image pair using a feature extraction network; Using guided filtering or conditional generation networks, the infrared temperature feature map is enhanced by using the visible light texture feature map as a condition to obtain an enhanced feature map; The enhanced feature map and the color components of the visible light image are fused to obtain the enhanced fused image.

4. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: Extracting the temperature distribution sub-matrix of each component in the enhanced fused image includes the following steps: The enhanced fused image is segmented into regions using a semantic segmentation network to obtain a classification label map, where each pixel in the classification label map has a corresponding device type identifier and component identifier; Based on the temperature grayscale calibration curve, the grayscale value of each pixel in the enhanced fused image is converted into a temperature value to obtain a temperature distribution matrix; Based on the spatial correspondence between the classification label map and the temperature distribution matrix, the temperature value of the corresponding pixel position of each component is extracted to form the temperature distribution submatrix of that component.

5. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: The thermal network model defines internal heat source nodes, external surface temperature nodes, and thermal resistance and thermal capacity connection relationships between nodes based on the actual structure of the equipment and the actual heat transfer path. The thermal resistance and thermal capacity connection relationships are based on the actual heat transfer path and utilize conduction units, convection units, and radiation units to connect each node. When calling the thermal network model corresponding to the substation equipment, the following steps are included: Mapping component identifiers identified from the enhanced fused image to nodes in the thermal network model; Load the default thermal resistance and thermal capacity parameter set in the thermal network model, and read the current equipment load rate and ambient temperature from the substation monitoring system or field sensors as boundary conditions.

6. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: The output of the thermal network model for the internal heat source distribution and additional loss values ​​of the device includes the following steps: The temperature distribution submatrix is ​​discretized along the surface of the device housing into several local temperature nodes, and the temperature value of each local temperature node is the average value of the pixel temperature in that area; The external surface temperature nodes in the thermal network model are expanded into several local external surface temperature nodes, each corresponding to a local temperature node; the internal heat sources of the equipment are divided into several internal heat source nodes according to their spatial location. Establish a thermal path connection between the internal heat source node and the local external surface temperature node. Based on the internal heat source node, the local external surface temperature node and the corresponding thermal path connection, establish a set of thermal network equations. The set of thermal network equations is to calculate the temperature value of each local external surface temperature node based on the assumed internal heat source node. An inversion objective function is established, which is the sum of squared deviations between the calculated temperature values ​​and the measured temperature values ​​of the local outer surface temperature nodes; The optimal solution for the internal heat source distribution that minimizes the inversion objective function is obtained based on the optimization algorithm. The additional loss value of all internal heat source nodes is obtained by summing the differences between the heat source intensity value of each internal heat source node in the optimal solution of the internal heat source distribution and the corresponding preset heat source intensity value.

7. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 6, characterized in that: The heat network equations are as follows: ； in: , For node indexing, node For nodes Adjacent nodes, traverse There are 1, 2, 3...M nodes; M=K+N, where M represents the number of nodes, K is the number of internal heat source nodes, and N is the number of local external surface temperature nodes. The internal heat source nodes and the local external surface temperature nodes are uniformly numbered as 1, 2, 3...M. For nodes Temperature; For nodes The heat capacity; For nodes The intensity of the heat source, for internal heat source nodes , The intensity of the heat source to be inverted; for local outer surface temperature nodes ,but ; Represents a node With nodes The equivalent thermal resistance between them; Represents nodes The set of directly connected nodes; Represents a node The sum of heat flows transmitted to all adjacent nodes.

8. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 1, characterized in that: The safety and energy efficiency index analysis based on the internal heat source distribution of the equipment and the additional loss value includes the following steps: Extract the regional distribution characteristics of heat sources and the anomaly degree of heat source intensity from the internal heat source distribution; Defect categories are identified based on the distribution characteristics of heat source areas, the anomaly degree of heat source intensity, and the temperature gradient vector to obtain the defect category and the defect severity level, and the corresponding safety score is determined based on the defect severity level; wherein, the temperature gradient vector is obtained by calculating the temperature gradient between the temperature distribution submatrix of the component and the spatial coordinate range of the corresponding component; The energy efficiency degradation rate is obtained by comparing the additional loss value with the rated power of the substation equipment; and the corresponding energy efficiency score is determined based on the energy efficiency degradation rate. The operating score of the substation equipment is obtained by weighting the safety score and the energy efficiency score.

9. The method for energy efficiency and safety testing of substation equipment based on thermal network inversion according to claim 5, characterized in that: It also includes building a modular thermal network model library for different equipment types; The modular thermal network model library comprises the following three-layer architecture: The base layer includes conduction units, convection units, and radiation units; each unit has standardized input / output interfaces; for example, the input is temperature difference / heat flux, and the output is heat flux / temperature difference; wherein, the conduction unit is: ； ； in, For conductive thermal resistance, This is the length of the heat transfer path. The thermal conductivity of the material. This refers to the heat transfer cross-sectional area; For heat capacity, For material density, For specific heat capacity, Unit volume; The convection unit is: ； in, For convective thermal resistance, The convective heat transfer coefficient is... For heat exchange area; The radiating element is: ； in The Stefan-Boltzmann constant is... For surface emissivity, For the radiation area, , These are the absolute temperatures of the radiating surface and the receiving surface, respectively. The equipment layer constructs a corresponding modular thermal network model based on the structure and heat transfer path of different types of equipment. The parameter layer is configured with a default set of thermal resistance and thermal capacity parameters for each modular thermal network model, and has an online calibration interface for model parameters; it is used to read the current equipment load rate and ambient temperature from the substation monitoring system or field sensors as boundary conditions. Before calling the thermal network model corresponding to the substation equipment, the corresponding thermal network model is selected from the modular thermal network model library based on the equipment identifier identified from the enhanced fusion image.

10. A substation equipment energy efficiency and safety monitoring system based on thermal network inversion, characterized in that: The method for energy efficiency and safety testing of substation equipment based on thermal network inversion as described in any one of claims 1-9 includes: The acquisition module is used to simultaneously acquire visible light and infrared thermal images of substation equipment; The registration module is used to spatially register the visible light image and the infrared thermal image to obtain a registered image pair; An enhancement module is used to enhance and fuse the registered image pairs to obtain an enhanced fused image; The extraction module is used to extract the temperature distribution sub-matrix of each component in the enhanced fused image; The inversion module is used to call the thermal network model corresponding to the substation equipment. The thermal network model takes the temperature distribution submatrix of each component as input and outputs the internal heat source distribution and additional loss value of the equipment. The analysis module is used to perform safety and energy efficiency index analysis based on the internal heat source distribution of the equipment and the additional loss value, so as to obtain the defect category and operating score of the substation equipment.

Images