X-ray image comparison and structure change analysis method for substation GIS equipment

Abstract

The application provides a substation GIS equipment-oriented X-ray image comparison and structure change analysis method, and relates to the technical field of power equipment detection.The method comprises the following steps: controlling a mobile detection device carrying an X-ray machine and a detector to perform first detection, collecting a first X-ray image, and recording reference multidimensional information of the mobile detection device to form a reference data set; comparing the real-time collected current multidimensional information with the reference data set, adjusting the device pose until the error meets a preset tolerance threshold; controlling the mobile detection device to collect a current X-ray image under the same equipment parameters, and performing image registration on the current X-ray image and the first X-ray image to obtain a registered current image; calculating a difference image between the current image and the first X-ray image, and extracting and quantifying a structure change region of the GIS equipment.The method can effectively improve the detection efficiency, ensure standard unification, and realize trend prediction.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, specifically to a method for X-ray image comparison and structural change analysis of substation GIS equipment. Background Technology

[0002] Gas-insulated switchgear (GIS) is a core piece of equipment in substations, and its structural integrity directly affects the safe and stable operation of the power grid. Due to long-term exposure to electrodynamic forces, thermal expansion and contraction, and mechanical vibrations, GIS equipment may experience subtle structural changes such as rod displacement, bolt loosening, and contact deformation. If these changes are not detected in time, they can easily lead to major malfunctions.

[0003] X-ray digital imaging technology has become an important tool for GIS equipment condition assessment due to its ability to penetrate equipment casings non-destructively and visually display internal structures. However, in practical applications, achieving accurate comparison of inspection results for the same equipment and the same part at different time points to identify minute structural changes remains a significant challenge.

[0004] The existing technology has the following main shortcomings: First, the repeatability of positioning is low, and the detection conditions are difficult to reproduce. Current detection procedures typically rely on manual experience or simple markings to roughly determine the detection location. Due to the complex environment of substations and the multiple degrees of freedom of the detection device itself (horizontal position, height, pitch angle, yaw angle, etc.), there are significant deviations in the relative positions and angles of the X-ray source, detector, and the device under test during each detection. This non-repeatability of the pose results in differences in viewing angle, magnification, and geometric deformation between two acquired images, introducing huge errors for subsequent direct image comparison.

[0005] Second, image registration accuracy is poor, and change detection sensitivity is insufficient. When poses are inconsistent, direct image difference analysis can introduce numerous "spurious changes" caused by viewpoint differences, thus masking genuine structural changes. Existing methods mostly employ registration algorithms based on the image content itself (such as feature points and mutual information). In scenarios like substation GIS, where texture features are sparse and structural similarity is high, such methods are prone to registration failures or misregistration, significantly reducing the reliability and sensitivity of change detection.

[0006] Third, there is a lack of quantitative methods for analyzing structural changes. Even if differences are found through manual comparison, current technologies lack automated quantitative indicators to objectively assess the severity of structural changes (such as the area ratio of the changed region). This makes the detection results heavily reliant on the subjective experience of analysts, making it difficult to achieve standardized and reproducible assessments of defect development trends.

[0007] Therefore, there is an urgent need for a method that can ensure highly consistent detection conditions, achieve high-precision automatic image registration, and perform quantitative analysis of structural changes, so as to improve the automation level and reliability of substation GIS equipment condition detection. Summary of the Invention

[0008] To address the problems in related technologies, this invention provides a method for X-ray image comparison and structural change analysis of substation GIS equipment.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Methods for X-ray image comparison and structural change analysis of substation GIS equipment include: Step S1: At the target GIS equipment detection point, control the mobile detection device equipped with an X-ray machine and detector to perform the first detection, acquire the first X-ray image, and simultaneously record the reference multidimensional information of the mobile detection device when acquiring the image to form a reference dataset. The reference multidimensional information includes reference pose information, equipment parameter information, and distance information between the detector and surrounding GIS equipment. Step S2: When repeated detection is required, import the benchmark dataset, guide the mobile detection device to move to the target GIS device location, compare the current multidimensional information collected in real time with the corresponding information in the benchmark dataset, and adjust the device pose until the error between the current multidimensional information and the benchmark multidimensional information meets the preset tolerance threshold. Step S3: After the device pose adjustment meets the requirements, control the moving detection device to acquire the current X-ray image under the same equipment parameters, and perform image registration with the first X-ray image in the reference dataset to obtain the registered current image; Step S4: Calculate the difference image between the registered current image and the initial X-ray image, and analyze and process the difference image to extract and quantify the structural change areas of the GIS equipment.

[0010] Optionally, in step S2, the adjustment of the device pose specifically includes: Based on the current pose vector corresponding to the current pose information from the real-time collected multi-dimensional information. The reference pose vector corresponding to the reference multidimensional information in the reference dataset Calculate the pose error vector E, and drive the motion detection device to move through the controller so that the norm of the pose error vector E is less than or equal to a preset error threshold. Where X and Y represent the X-axis and Y-axis coordinates of the moving detection device in the preset coordinate system during the current detection, respectively, and Z represents the lifting height of the moving detection device during the current detection. , and These represent the yaw angle, pitch angle, and roll angle of the moving detection device, respectively. and These represent the X-axis and Y-axis coordinates of the moving detection device in the preset coordinate system during the initial detection. This indicates the lifting height of the moving detection device during the initial test. , and These represent the yaw angle, pitch angle, and roll angle of the moving detection device during the initial test, respectively.

[0011] Optionally, in step S3, the image registration includes an initial transformation sub-step and a fine registration sub-step; The initial transformation sub-step is based on the pose error vector E calculated in step S2, utilizing the horizontal position deviation. and yaw angle deviation Construct an initial affine transformation matrix T, where, This indicates the positional deviation of the position detection device in the X-axis direction within the horizontal plane, i.e. , This indicates the positional deviation of the position detection device in the Y-axis direction within the horizontal plane, i.e. , The angular deviation of the yaw angle of the position detection device, i.e. ; The fine registration sub-step optimizes the initial affine transformation matrix T based on an image feature point matching algorithm to obtain the optimized transformation matrix. and utilize Transform the current X-ray image.

[0012] Optionally, the initial affine transformation matrix T is calculated using the following formula: In the formula, cos and sin represent the cosine trigonometric function and the sine trigonometric function, respectively.

[0013] Optionally, the fine registration sub-step specifically includes: Feature point sets are extracted from the first X-ray image and the current X-ray image using either the SIFT or ORB algorithm. For F, feature points are matched using the RANSAC algorithm, and mismatched point pairs are eliminated. Based on the correct matching points, the initial affine transformation matrix T is optimized to obtain the optimized transformation matrix. .

[0014] Optionally, in step S4, the calculation process of the difference image specifically includes: Registered current image With the first X-ray image Perform pixel-by-pixel grayscale value difference and take the absolute value to obtain the difference image D, i.e.: In the formula, x and y represent the two-dimensional coordinates of a pixel in the image. Represents the first X-ray image in the benchmark dataset. In coordinates The pixel grayscale value at that location, This represents the current X-ray image after registration transformation. In coordinates The pixel grayscale value at that location.

[0015] Optionally, the extraction and quantification of the structural change areas of the GIS equipment specifically includes: An adaptive threshold binarization process is performed on the difference image D to obtain a binarized change region image R, wherein the adaptive threshold is determined based on the local statistical characteristics of the difference image D. Calculate the area ratio of the effective change region in the binarized change region image R. , as a quantitative indicator of structural change.

[0016] Optionally, the binarized change region image R is obtained by the following formula: In the formula, This represents the adaptive threshold used for binarization. The area percentage of the effective change region Calculated using the following formula: In the formula, This represents summing all pixel values ​​in the binarized image R, where N is the total number of pixels in the image.

[0017] Optionally, in step S1, the motion detection device is equipped with a positioning module, an attitude sensor, and a distance measurement sensor. The positioning module is configured as a GNSS module supporting RTK technology, the attitude sensor is configured as an inertial measurement unit including a three-axis gyroscope and a three-axis accelerometer, and the distance measurement sensor is configured as a laser rangefinder.

[0018] Optionally, the method for X-ray image comparison and structural change analysis of substation GIS equipment further includes: Step S5: Compare the structural change index corresponding to the quantified GIS equipment structural change area with the preset safety threshold. If the index exceeds the safety threshold, automatically generate detection information containing the location of the change area and the corresponding quantification result.

[0019] Beneficial effects: 1. Through the above technical solution, firstly, the method of the present invention can achieve a high degree of reproducibility of detection conditions and provide a solid foundation for accurate comparison. Specifically, steps S1 and S2 constitute a record-reproducibility loop. In step S1, the baseline multidimensional information at the time of the first detection is systematically recorded. In step S2, during repeated detection, the active guidance device compares the current multidimensional information collected in real time with the baseline multidimensional information and adjusts the pose to ensure that the error meets the preset tolerance threshold. In this way, not only can subjective errors be effectively eliminated, but also detection consistency can be effectively ensured.

[0020] Second, the method of this invention can effectively improve the accuracy and reliability of image comparison and effectively suppress spurious changes. Specifically, although step S2 strives to ensure pose consistency, under extremely high precision requirements, minute residual deviations may still cause pixel-level misalignment between the two images. Step S3 of this invention introduces image registration as a key step to actively compensate for and align these minute, unavoidable residual deviations digitally. In this way, through registration, image differences caused by geometric factors such as viewpoint, translation, and rotation can be eliminated to the greatest extent possible, thereby highlighting image differences truly caused by structural changes and improving the signal-to-noise ratio.

[0021] Third, the method of this invention enables objective and quantitative assessment of structural changes, enhancing the authority and usability of the detection results. Specifically, step S4 involves calculating and analyzing a differential image, with the ultimate goal of extracting and quantifying the area of ​​structural change. Thus, the output is no longer a qualitative judgment relying on personal experience, but an objective and quantifiable indicator, making the detection results recordable, traceable, and comparable. Simultaneously, the quantified results provide direct and objective data support for equipment condition assessment, risk level classification, and maintenance decision-making.

[0022] Overall, the method of this invention can achieve an automated, standardized, and intelligent closed-loop detection process. Specifically, executing steps S1 to S4 sequentially forms a complete automated process: recording the baseline - automatic navigation and adjustment - automatic registration - automatic analysis and quantification. This not only effectively improves detection efficiency but also ensures standardization and enables trend prediction.

[0023] 2. Other beneficial effects or advantages of the present invention will be described in detail in the specific embodiments. Attached Figure Description

[0024] 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 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 these drawings without creative effort.

[0025] in: Figure 1 This is a flowchart illustrating the steps of an exemplary embodiment of the present invention for a method of X-ray image comparison and structural change analysis of substation GIS equipment; Figures 2 to 4 These are lifting mechanisms provided in an exemplary embodiment of the present invention (wherein, a rotating structure and an X-ray machine are also shown), wherein, Figure 2 Corresponding to the independent rotation (pitch) large lift device (0-3.5 meters adjustable, angle ±90 degrees); Figure 3 Corresponding to item 2, a separate rotation (pitch) small lifting device (adjustable from 0-1.5 meters, angle ±90 degrees); Figure 4 Corresponding to 3, Individual rotation (pitch) - medium lifting device (0-2.5 meters adjustable, angle ±90 degrees). Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0027] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0028] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. It should also be noted that in embodiments of this invention, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in embodiments of this invention should not be construed as preferred or advantageous over other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0029] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0030] like Figure 1 As shown, this embodiment provides a method for X-ray image comparison and structural change analysis of substation GIS equipment, including: Step S1: At the target GIS equipment detection point, control the mobile detection device equipped with an X-ray machine and detector to perform the first detection, acquire the first X-ray image, and simultaneously record the reference multidimensional information of the mobile detection device when acquiring the image to form a reference dataset. The reference multidimensional information includes reference pose information, equipment parameter information, and distance information between the detector and surrounding GIS equipment. Step S2: When repeated detection is required, import the benchmark dataset, guide the mobile detection device to the target GIS device location, compare the current multidimensional information collected in real time with the corresponding information in the benchmark dataset, and adjust the device pose until the error between the current multidimensional information and the benchmark multidimensional information meets the preset tolerance threshold. Step S3: After the device pose adjustment meets the requirements, control the moving detection device to acquire the current X-ray image under the same equipment parameters, and perform image registration with the first X-ray image in the reference dataset to obtain the registered current image; Step S4: Calculate the difference image between the registered current image and the initial X-ray image, analyze and process the difference image, and extract and quantify the structural change areas of the GIS equipment.

[0031] Through the above technical solution, firstly, the method of the present invention can achieve high reproducibility of detection conditions and provide a solid foundation for accurate comparison. Specifically, steps S1 and S2 constitute a record-reproducibility loop. In step S1, the reference multidimensional information (including pose, equipment parameters, and distance information) at the time of the first detection is systematically recorded. In step S2, during repeated detection, the active guidance device compares the current multidimensional information collected in real time with the reference multidimensional information and adjusts the pose to ensure that the error meets the preset tolerance threshold. In this way, not only can subjective errors be effectively eliminated (avoiding the problem of inaccurate positioning caused by different operators or the same operator at different times due to differences in judgment), but also detection consistency can be effectively ensured (making the spatial geometric relationship and equipment working status between the X-ray machine, detector, and the tested GIS equipment highly consistent in the two detections, so that the current X-ray image and the first X-ray image are comparable, that is, the image difference mainly comes from the structural changes of the equipment itself, rather than the interference introduced by inconsistent detection conditions).

[0032] Second, the method of this invention can effectively improve the accuracy and reliability of image comparison and effectively suppress spurious changes. Specifically, although step S2 strives to ensure pose consistency, under extremely high precision requirements, minute residual deviations may still cause pixel-level misalignment between the two images. Step S3 of this invention introduces image registration as a key step to actively compensate and align these minute, unavoidable residual deviations digitally. In this way, through registration, image differences caused by geometric factors such as viewpoint, translation, and rotation (these are noise or spurious changes) can be eliminated to the greatest extent possible, thereby highlighting image differences truly caused by structural changes and improving the signal-to-noise ratio.

[0033] Third, the method of this invention enables objective and quantitative assessment of structural changes, enhancing the authority and usability of the detection results. Specifically, step S4 involves calculating and analyzing a differential image, with the ultimate goal of extracting and quantifying the area of ​​structural change. Thus, the output is no longer a qualitative judgment relying on personal experience, but an objective and quantifiable indicator, making the detection results recordable, traceable, and comparable. Simultaneously, the quantified results provide direct and objective data support for equipment condition assessment, risk level classification, and maintenance decision-making.

[0034] Overall, the method of this invention achieves an automated, standardized, and intelligent closed-loop detection process. Specifically, executing steps S1 to S4 sequentially forms a complete automated process: recording baseline - automatic navigation and adjustment - automatic registration - automatic analysis and quantification. This not only effectively improves detection efficiency (effectively reducing reliance on highly skilled operators, minimizing human intervention, and significantly enhancing detection efficiency), but also ensures standardized processes (the entire process is controlled by the system, ensuring that detections performed at different times and by different personnel follow the same standard, guaranteeing data consistency and comparability). Furthermore, it enables trend prediction (since the results of each detection are quantifiable, long-term accumulated data can be used to analyze trends in GIS equipment structural changes, providing the possibility for predictive maintenance).

[0035] The method of the present invention will be described below with reference to a specific embodiment.

[0036] Scene description: Periodic inspections were conducted on the "A-phase busbar cylinder flange connection" of a GIS equipment in a 500kV substation to monitor for possible displacement or loosening of the internal guide rod.

[0037] Step 1: Initial test data collection.

[0038] 1. Device configuration.

[0039] The mobile detection device is equipped with the following hardware: Positioning module: A GNSS module that supports RTK (such as WTRTK-980) used to obtain planar coordinates (X,Y) with centimeter-level accuracy.

[0040] Attitude sensor: IMU (such as JY901S), used to measure yaw angle (θ), pitch angle (φ), and roll angle (ψ).

[0041] Distance measurement sensor: Laser rangefinder (such as Keyence IL-300), used to measure the distance vectors from the four sides of the detector to surrounding GIS equipment. .

[0042] Lifting mechanism (such as) Figures 2 to 4 As shown): Equipped with an encoder, it can accurately measure the lifting height (Z).

[0043] X-ray machine and detector: parameters can be precisely controlled.

[0044] 2. Data Collection: The operator moves the device to the optimal testing point.

[0045] The control device performed the detection and acquired a clear first X-ray image. .

[0046] Simultaneously record benchmark multidimensional information to form benchmark dataset B: Reference pose information : =[100.500m,50.800m,1.200m,45.5°,-0.5°,0.3°].

[0047] Equipment parameter information : =[Tube voltage = 220kV, tube current = 8.0mA, exposure time = 0.8s].

[0048] Distance information : =[1.05m, 1.02m, 0.98m, 1.01m].

[0049] Step 2: Repeat the detection and positioning adjustment.

[0050] Repeat the test after a period of time (e.g., three months later).

[0051] 1. Import the baseline dataset B, and the system plans the path guidance device to move to the vicinity of the target GIS device.

[0052] 2. Real-time acquisition of current multidimensional information to obtain the current pose vector P: P=[100.508m,50.795m, 1.185m,45.0°,-0.6°,0.35°].

[0053] Calculate the pose error vector E: =100.508-100.500=0.008m; = 50.795 - 50.800 = -0.005m; =1.185-1.200=-0.015m; = 45.0° - 45.5° = -0.5°; =-0.6°-(-0.5°)=-0.1°; =0.35°-0.3°=0.05°.

[0054] Therefore, the error vector 3. Adjusting the device's pose: The system calculates the error norm. And fine-tuning is achieved through a PID controller drive device. The adjustment process continues until... ( The error threshold and The final adjusted P is [100.500m, 50.800m, 1.200m, 45.5°, -0.5°, 0.3°], and the error meets the tolerance threshold.

[0055] Step 3: Image acquisition and registration.

[0056] 1. Image Acquisition: After the pose adjustment meets the requirements, control the X-ray machine to... The current X-ray image I was acquired using the exact same parameters (220kV, 8.0mA, 0.8s).

[0057] 2. Image registration: Initial transformation: Calculate the deviation using the adjusted final pose: =0m, =0m, =0°.

[0058] Construct the initial affine transformation matrix T: Among them, the angle needs to be converted to radians for calculation, cos(0°)=1, sin(0°)=0; Fine-grained registration: Using the SIFT algorithm from Feature point sets are extracted from I and I respectively. And F.

[0059] Feature point matching is performed using the RANSAC algorithm to eliminate incorrect matching pairs.

[0060] Based on the correct matching point pairs, the initial transformation matrix T is optimized to obtain the optimized transformation matrix. . It can more accurately compensate for minor residual deformations between images.

[0061] use Transform the current image I to obtain the same as The current image after strict alignment and registration .

[0062] Step 4: Change detection and quantitative analysis.

[0063] Calculate the difference image D: right and Perform pixel-by-pixel grayscale value difference and take the absolute value: .

[0064] The difference image D is obtained, where bright spots represent regions where changes may occur.

[0065] Extract and quantify the regions of change: Adaptive threshold binarization: Calculate the adaptive threshold based on the local statistical properties of the difference image D (e.g., using the Otsu algorithm or based on local mean and standard deviation). Assuming the calculation yields... =15 (grayscale value range 0-255).

[0066] Generate a binarized image R: Calculation of area ratio : Let the total number of pixels in the image be N = 1024 × 1024 = 1048576.

[0067] Count the total number of pixels with a value of 1 in a binary image R: =1500.

[0068] Calculate the area percentage of the effective change region: =1500 / 1048576 is approximately 0.14%.

[0069] Step 5: Result Generation and Output.

[0070] The system will quantify the structural change indicators =0.12% is compared with a preset safety threshold δ (e.g., δ=0.10%).

[0071] because (0.12%) > δ(0.10%), indicating that the system has a significant structural change.

[0072] An automatic detection report is generated, which includes: detection location, baseline and current images, difference image, changed area identifiers, and percentage of changed area. The system will then issue an alarm message, prompting maintenance personnel to pay close attention to and verify the "A-phase busbar cylinder flange connection".

[0073] In the above embodiments, it should be noted that, firstly, regarding the calculation formula for the initial affine transformation matrix T, although the formula itself is a standard formula in computer graphics, its application in this application has a special improvement point, namely, the innovation of the parameter source. The initial transformation of traditional image registration is guessed or estimated based on low-level image features, while in this solution, , and This is directly derived from physical measurements taken from high-precision positioning systems (such as RTK GNSS) and attitude sensors (IMUs). It serves as a crucial bridge combining precise geometric measurements of the physical world with image processing. This initial transformation based on physical measurements effectively improves the convergence speed and success rate of subsequent image registration. It provides an initial value that closely approximates reality for fine registration, avoiding the problems of traditional methods getting stuck in local optima or failing to register due to excessive initial value deviations. This improvement is particularly important in scenarios like substation GIS where texture features are not readily apparent.

[0074] Secondly, the calculation formula for the difference image D is the simplest image differencing algorithm. It is precisely because of its simplicity that, combined with the steps of this application, more efficient difference calculation can be achieved. Specifically, the difference operation in this scheme is performed on two images that have undergone "high-precision pose reproduction" and "fine registration based on physical measurement initialization." This minimizes spurious changes caused by inconsistent detection conditions (viewpoint, position). Because most of the interference factors are eliminated, this simple difference formula can efficiently and clearly highlight the true structural changes. If this formula is used directly on an unregistered image, the result will be noisy and unrecognizable. Therefore, the improvement of this formula mainly lies in constructing an advanced preprocessing workflow that maximizes the efficiency of this simple formula.

[0075] Third, regarding the calculation formula for the binarized change region image R, the main improvement of this invention lies in the adaptive threshold. The method for determining the threshold. Specifically, in this invention, an adaptive threshold is used. The threshold is determined based on the local statistical characteristics of the difference image D. This is a significant improvement over using a fixed threshold, which cannot adapt to changes in brightness, contrast, and noise levels in different regions of the image. Adaptive thresholding (such as the Otsu method or methods based on local mean and standard deviation) can dynamically calculate the optimal threshold τ based on the statistical distribution of the D image itself. This allows for a more accurate separation of real changes from residual noise, effectively improving the robustness and accuracy of change region extraction, and avoiding the problems of fixed thresholds potentially missing detections in dark areas and misclassifying noise as changes in bright areas.

[0076] For example, adaptive threshold The process of determining can be as follows: Objective: To compute an optimal global threshold for the difference image D. This is used to categorize pixels into two types: "changed" and "unchanged". Input: Gray-level difference image D, whose pixel gray-level values ​​range from [0, L-1] (e.g., for an 8-bit image, L=256); 1. Calculate the grayscale histogram: Each gray level in the statistical difference image D The number of pixels that appear is used to obtain a histogram. .

[0077] Calculate the total number of pixels N: N = .

[0078] 2. Calculate the normalized histogram (probability distribution): Calculate each gray level The probability of occurrence is p(i): p(i) = hist[i] / N.

[0079] 3. Iterate through all possible thresholds t: Assume that a threshold t divides pixels into two classes, C0 and C1: C0: Pixels with gray values ​​in [0, t] (considered as "background" or "unchanged" areas).

[0080] C1: Pixels with gray values ​​in [t+1, L-1] (considered as "foreground" or "variable" regions).

[0081] 4. For each candidate threshold t, calculate the inter-class variance. : Calculate the cumulative probability ω0(t) of class C0: ω0(t) = Σp(i) (i from 0 to t).

[0082] Calculate the average gray level μ0(t) of class C0: μ0(t)=[Σi*p(i)] / ω0(t) (i from 0 to t).

[0083] Calculate the cumulative probability ω1(t) of class C1: ω1(t) = 1 - ω0(t).

[0084] Calculate the average gray level μ1(t) of class C1: μ1(t) = [Σi*p(i)] / ω1(t) (i ranges from t+1 to L-1). Alternatively, it can be calculated using the global average gray level μ_T: μ1(t) = (μ_T - ω0(t) * μ0(t)) / ω1(t), where μ_T = Σi*p(i) (i ranges from 0 to L-1).

[0085] Calculate the between-class variance : =ω0(t)*ω1(t)*[μ0(t)-μ1(t)]²; 5. Find the optimal threshold τ: Iterate through all possible values ​​of t (from 0 to L-1) and find the t that maximizes the inter-class variance; this is the adaptive threshold. .

[0086] Fourth, the area proportion of the effective change zone. In terms of the calculation formula, it provides an innovation from pixels to engineering parameters: the ultimate goal of this scheme is not to produce an image marked with change points, but to provide an objective and comparable indicator of structural change. It allows for a leap from qualitative (there is change) to quantitative (how much has changed). This makes historical trend analysis possible (e.g., three consecutive tests). The percentages are 0.05%, 0.08%, and 0.15%, respectively, indicating that the defect is developing. Furthermore, this makes it possible to set a unified alarm threshold (e.g., δ = 0.10%), achieving standardization and automation of detection result judgment and reducing reliance on human experience. In addition, it can provide core data support for predictive maintenance.

[0087] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for X-ray image comparison and structural change analysis of substation GIS equipment, characterized in that, include: Step S1: At the target GIS equipment detection point, control the mobile detection device equipped with an X-ray machine and detector to perform the first detection, acquire the first X-ray image, and simultaneously record the reference multidimensional information of the mobile detection device when acquiring the image to form a reference dataset. The reference multidimensional information includes reference pose information, equipment parameter information, and distance information between the detector and surrounding GIS equipment. Step S2: When repeated detection is required, import the benchmark dataset, guide the mobile detection device to move to the target GIS device location, compare the current multidimensional information collected in real time with the corresponding information in the benchmark dataset, and adjust the device pose until the error between the current multidimensional information and the benchmark multidimensional information meets the preset tolerance threshold. Step S3: After the device pose adjustment meets the requirements, control the moving detection device to acquire the current X-ray image under the same equipment parameters, and perform image registration with the first X-ray image in the reference dataset to obtain the registered current image; Step S4: Calculate the difference image between the registered current image and the initial X-ray image, and analyze and process the difference image to extract and quantify the structural change areas of the GIS equipment; In step S2, the adjustment device pose specifically includes: Based on the current pose vector corresponding to the current pose information from the real-time collected multi-dimensional information. The reference pose vector corresponding to the reference multidimensional information in the reference dataset Calculate the pose error vector E, and drive the motion detection device to move through the controller so that the norm of the pose error vector E is less than or equal to a preset error threshold. Where X and Y represent the X-axis and Y-axis coordinates of the moving detection device in the preset coordinate system during the current detection, respectively, and Z represents the lifting height of the moving detection device during the current detection. , and These represent the yaw angle, pitch angle, and roll angle of the moving detection device, respectively. and These represent the X-axis and Y-axis coordinates of the moving detection device in the preset coordinate system during the initial detection. This indicates the lifting height of the moving detection device during the initial test. , and These represent the yaw angle, pitch angle, and roll angle of the moving detection device during the initial test, respectively. In step S3, the image registration includes an initial transformation sub-step and a fine registration sub-step; The initial transformation sub-step is based on the pose error vector E calculated in step S2, utilizing the horizontal position deviation. and yaw angle deviation Construct an initial affine transformation matrix T, where, This indicates the positional deviation of the position detection device in the X-axis direction within the horizontal plane, i.e. , This indicates the positional deviation of the position detection device in the Y-axis direction within the horizontal plane, i.e. , The angular deviation of the yaw angle of the position detection device, i.e. ; The fine registration sub-step optimizes the initial affine transformation matrix T based on an image feature point matching algorithm to obtain the optimized transformation matrix. and utilize Transform the current X-ray image; The fine registration sub-step specifically includes: Feature point sets are extracted from the first X-ray image and the current X-ray image using either the SIFT or ORB algorithm. For F, feature points are matched using the RANSAC algorithm, and mismatched point pairs are eliminated. Based on the correct matching points, the initial affine transformation matrix T is optimized to obtain the optimized transformation matrix. .

2. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to claim 1, characterized in that, The initial affine transformation matrix T is calculated using the following formula: In the formula, cos and sin represent the cosine trigonometric function and the sine trigonometric function, respectively.

3. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to claim 1, characterized in that, In step S4, the calculation process of the difference image specifically includes: Registered current image With the first X-ray image Perform pixel-by-pixel grayscale value difference and take the absolute value to obtain the difference image D, i.e.: In the formula, x and y represent the two-dimensional coordinates of a pixel in the image. Represents the first X-ray image in the benchmark dataset. In coordinates The pixel grayscale value at that location, This represents the current X-ray image after registration transformation. In coordinates The pixel grayscale value at that location.

4. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to claim 3, characterized in that, The extraction and quantification of structural change regions of GIS equipment specifically includes: An adaptive threshold binarization process is performed on the difference image D to obtain a binarized change region image R, wherein the adaptive threshold is determined based on the local statistical characteristics of the difference image D. Calculate the area ratio of the effective change region in the binarized change region image R. , as a quantitative indicator of structural change.

5. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to claim 4, characterized in that, The binarized change region image R is obtained by the following formula: In the formula, This represents the adaptive threshold used for binarization. The area percentage of the effective change region Calculated using the following formula: In the formula, This represents summing all pixel values ​​in the binarized image R, where N is the total number of pixels in the image.

6. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to claim 1, characterized in that, In step S1, the motion detection device is equipped with a positioning module, an attitude sensor, and a distance measurement sensor. The positioning module is configured as a GNSS module supporting RTK technology, the attitude sensor is configured as an inertial measurement unit including a three-axis gyroscope and a three-axis accelerometer, and the distance measurement sensor is configured as a laser rangefinder.

7. The method for X-ray image comparison and structural change analysis of substation GIS equipment according to any one of claims 1-6, characterized in that, The X-ray image comparison and structural change analysis method for substation GIS equipment also includes: Step S5: Compare the structural change index corresponding to the quantified GIS equipment structural change area with the preset safety threshold. If the index exceeds the safety threshold, automatically generate detection information containing the location of the change area and the corresponding quantification result.

Images