Cable fire resistance test method and system based on machine vision and image processing

Abstract

The invention discloses a cable fire resistance test method and system based on machine vision and image processing, and particularly relates to the technical field of cable tests.The cable fire resistance test method comprises the steps that system construction and thermal stability calibration are conducted, data are synchronously collected through multispectral binocular vision and a thermal infrared imager, and thermal calibration errors are compensated online through a temperature-parameter drift model; multispectral image processing and target extraction are carried out, flames are inhibited through a near-infrared band, and the cable contour is separated by adopting a frequency domain filtering technology; performing three-dimensional deformation measurement and thermal correction, reconstructing a three-dimensional point cloud in combination with stereo matching and confidence filtering, and deducting a deformation component caused by temperature based on a thermal expansion physical model to obtain a real mechanical deformation parameter; and performing multi-modal data fusion and failure judgment, and synchronously fusing deformation, structure, temperature and circuit state data. According to the invention, flame interference and high-temperature environment influence are effectively overcome, and high-precision automatic measurement and intelligent judgment of multiple parameters such as deformation and ablation in the cable fire-resistant process are realized.

Description

Technical Field

[0001] This invention relates to the field of cable testing technology, and more specifically, to a method and system for testing the fire resistance of cables based on machine vision and image processing. Background Technology

[0002] As a critical carrier for power transmission and signal control, the fire resistance performance of cables directly affects the safety of life and property and the continued operation of critical facilities in the event of a fire. Therefore, standardized fire resistance testing of cables to assess their ability to maintain circuit integrity under specified flame conditions is of paramount engineering significance. Currently, the fire resistance testing methods specified in domestic and international standards mainly rely on continuously applying a flame in a dedicated combustion device and judging fire resistance performance by monitoring whether the circuit at both ends of the sample remains open or closed. This method is essentially a binary "pass / fail" judgment, which can only provide the final time point of failure and cannot reflect the evolution details of progressive damage such as ablation, deformation, and carbonization of the sheath and insulation layers during flame burning. The test results are highly dependent on manual observation and recording, and have inherent defects such as strong subjectivity, poor repeatability, and inability to obtain quantitative damage process data, making it difficult to meet the needs of in-depth research on the fire resistance mechanism of cables and refined comparative evaluation of product performance.

[0003] To overcome the limitations of manual judgment, existing technologies have introduced machine vision-based observation methods. A common approach is to use visible light cameras to monitor the testing process and evaluate flame morphology or cable appearance changes through image analysis. However, in real fire resistance testing environments, intense open flames, dense smoke, and high-temperature heat radiation and heat wave fluctuations pose severe challenges to visible light-based imaging systems. The high brightness and dynamic texture of the flame itself can severely interfere with or even completely obscure the edge contours of the cable, rendering traditional image segmentation algorithms ineffective. High temperatures cause changes in the air refractive index and thermal deformation of the camera lens itself, introducing significant geometric distortion in the image, leading to systematic biases in two-dimensional or three-dimensional measurement results. Furthermore, existing vision methods often focus on two-dimensional information from a single viewpoint or a single spectrum, lacking the ability to accurately quantify the three-dimensional deformation of the cable (especially the unexposed side), and are unable to effectively distinguish between thermal expansion of materials and actual structural deformation, resulting in insufficient physical accuracy of the measurement results.

[0004] Some improved technologies attempt to incorporate infrared thermal imagers to monitor temperature or combine multiple sensors. However, these methods often merely present data in parallel, lacking effective mechanisms for synchronizing, fusing, and correlating multimodal information. For example, they fail to deeply integrate visible light ablation morphology, near-infrared structural information, temperature field distribution, and electrical performance indicators under a unified spatiotemporal reference, thus failing to construct a comprehensive quantitative index reflecting the decay of cable fire resistance. Furthermore, existing technologies generally lack effective online monitoring and compensation methods for the key error source—parameter drift of the imaging system itself caused by high-temperature environments (thermal calibration error)—limiting long-term measurement accuracy. Therefore, developing a cable fire resistance testing method and system capable of resisting strong flame interference, achieving accurate three-dimensional deformation measurement under high-temperature environments, and integrating multi-source information for intelligent failure determination has become an urgent technical problem to be solved in this field. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for testing the fire resistance of cables based on machine vision and image processing.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The cable fire resistance testing method based on machine vision and image processing includes the following steps:

[0008] B1. System Construction and Thermal Stability Calibration: A synchronous acquisition system including a visible-near-infrared binocular vision system, an infrared thermal imager, and a circuit monitoring device was built. The binocular system was calibrated, and online thermal stability compensation was performed based on the temperature-parameter drift model.

[0009] B2. Multispectral Image Processing and Target Extraction: Acquire visible light and near-infrared images, extract the cable target contour and generate a binary mask sequence through flame suppression, frequency domain filtering and morphological processing;

[0010] B3. Three-dimensional deformation measurement and thermal correction: Based on stereo matching of binocular images, the three-dimensional point cloud of the cable surface is reconstructed. Combined with infrared temperature data and the material's thermal expansion coefficient, the thermal expansion component is subtracted to obtain the true three-dimensional deformation parameters.

[0011] B4. Multimodal data fusion and failure determination: Simultaneously fuse deformation, ablation area, structural continuity, temperature and circuit status data, calculate comprehensive characteristic indicators through time-varying weighted fusion algorithm, and automatically determine the critical point and mode of fire resistance failure based on multiple criteria.

[0012] Specifically, the online thermal stability compensation in B1 includes:

[0013] Before testing, images of the calibration plate at different temperatures were acquired, the rate of change of key calibration parameters with temperature was calculated, and the camera intrinsic parameters were dynamically corrected based on the real-time monitored lens ambient temperature during the test. Key calibration parameters include principal point coordinates, focal length, and lens distortion coefficient.

[0014] Specifically, in B2:

[0015] Near-infrared narrowband filters of a specific wavelength band are used to suppress flame radiation, and a frequency domain low-pass filter is used to separate the high-frequency texture of flame jumping from the low-frequency information of cable outline.

[0016] Specifically, in B3:

[0017] The stereo matching algorithm employs a semi-global block matching algorithm and introduces a comprehensive confidence evaluation mechanism based on matching cost and normalized cross-correlation to filter low-confidence 3D points.

[0018] Specifically, in B3:

[0019] The temperature-expansion mapping model calculates the coordinate offset caused by thermal expansion based on the thermal expansion coefficients of each layer of cable material and infrared temperature measurement data, and deducts it from the total displacement.

[0020] Specifically, in B4:

[0021] The structural continuity index is calculated by extracting the cable skeleton and detecting break points; the fusion feature index uses a time-varying weight vector to weight and fuse data from different stages.

[0022] The cable fire resistance testing system based on machine vision and image processing includes the following modules:

[0023] The multispectral synchronous acquisition and high-precision calibration module is used to synchronously control the visible-near-infrared binocular camera and the infrared thermal imager, and to perform online thermal drift compensation.

[0024] An anti-interference cable image enhancement and segmentation module is used to suppress flame interference, extract cable contours, and generate a binary mask sequence.

[0025] The 3D thermal deformation measurement and physical correction module is used to reconstruct the 3D point cloud of the cable and output the true deformation parameters by subtracting the thermal expansion component based on the temperature data.

[0026] The multi-source information fusion and intelligent failure determination module is used to fuse multi-dimensional data streams and automatically determine fire resistance failure through time-varying weighted fusion and multiple criteria.

[0027] The technical effects and advantages of this invention are as follows:

[0028] This significantly improves the accuracy, stability, and automation of key parameter measurements in fire resistance testing. By employing multispectral imaging and frequency domain filtering techniques, it effectively overcomes the interference of intense flames and smoke on cable target identification, achieving robust extraction of cable contours in complex environments. The thermal stability calibration compensation mechanism and the physical model-based thermal expansion correction method fundamentally suppress image distortion and measurement drift caused by high-temperature environments, enabling high-precision quantification of the cable's three-dimensional deformation and ablation process. This solves the fundamental problems of traditional visual inspection methods and single visible light vision methods being greatly affected by environmental interference and resulting in distorted measurement results.

[0029] This system achieves multi-dimensional, intelligent, and objective assessment of fire resistance performance. By simultaneously integrating multi-modal data such as deformation, structure, temperature, and electrical parameters, and designing a time-varying fusion algorithm that reflects the failure evolution law, a comprehensive evaluation system has been constructed. It can automatically and accurately determine the failure critical point and mode based on multiple intelligent criteria such as abrupt deformation and structural fracture, overcoming the lag and randomness of single circuit continuity criteria. Finally, it automatically generates a digital report containing the complete evolution process, greatly reducing the reliance on human experience in the testing process and improving the reliability, repeatability, and comparability of test results. Attached Figure Description

[0030] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] like Figure 1 As shown, the steps of the cable fire resistance testing method based on machine vision and image processing are as follows:

[0033] Step 1: System Construction and Synchronous Data Acquisition. A synchronous acquisition system consisting of a visible-near-infrared binocular vision system, an infrared thermal imager, and a circuit monitoring device was constructed, establishing a high-precision, thermally interference-resistant measurement benchmark. An online thermal stability compensation calibration mechanism based on a temperature-parameter drift model was introduced. Through preheating calibration and linear modeling, camera intrinsic parameter drift caused by high-temperature environments was corrected in real time, ensuring the long-term stability and accuracy of three-dimensional geometric measurements throughout the flame test process. The process is as follows:

[0034] A fire resistance testing platform was constructed, and two synchronously triggered high-resolution multispectral industrial cameras (one equipped with a visible light filter and the other with a near-infrared narrowband filter with a specific center wavelength) were symmetrically arranged on both sides of the burner to form a binocular vision system. An infrared thermal imager was placed on the unexposed side of the cable.

[0035] All imaging devices and power monitoring units are connected to the central controller and receive the same trigger signal to begin synchronous acquisition. Before testing, the binocular vision system is calibrated to obtain intrinsic and extrinsic parameters and distortion coefficients.

[0036] A high-temperature resistant speckle pattern is pre-sprayed onto the surface of the cable sample as deformation tracking feature points. The combustion test is then initiated, simultaneously triggering all sensors to begin recording timing data.

[0037] To eliminate the problem of calibration parameter (internal and external parameters) drift caused by thermal deformation of the camera lens due to high-temperature radiation in the combustion chamber during testing, an optimization mechanism was used after completing the routine room temperature calibration; the specific process is as follows:

[0038] Preheating the calibration plate: Fix the calibration plate at the cable installation location in the test area. Start the burner or auxiliary heating device to allow the surface temperature of the calibration plate to rise uniformly and stabilize at a predicted intermediate temperature within the test environment. (For example Subsequently, the binocular system quickly acquired a set of calibration plate images at this temperature.

[0039] Calculation of thermally induced parameter drift: Solve for room temperature separately and intermediate temperature The two sets of camera intrinsic parameter matrices K and lens distortion coefficients are as follows Calculate the coordinates of the principal point. and the changes in key parameters such as focal length f : Where p is a key calibration parameter of the camera. ; For intermediate temperature The calibration parameter values ​​calculated at temperatures below (e.g., 80℃); To be at room temperature The calibration parameter values ​​calculated at (e.g., 25℃).

[0040] Establish a drift compensation model: Assume key calibration parameters With temperature The change in temperature is linear within the test temperature range; calculate its rate of temperature change. : For any time t during the test, based on the real-time monitored ambient temperature near the lens... It can perform online compensation on the original room temperature calibration parameters to obtain the corrected parameters. : ;in This is the rate of temperature change of the parameter. These are the initial values ​​for the calibration parameters at room temperature. For testing time The ambient temperature near the lens is monitored in real time;

[0041] Real-time application: In the subsequent image processing and 3D reconstruction steps two and three, the camera parameters, which are corrected in real time according to the model, are used instead of the fixed room temperature calibration parameters to offset the systematic errors introduced by thermal deformation and ensure the accuracy of geometric measurements.

[0042] Step Two: Multispectral Image Processing and Target Extraction. The cable target contour is robustly separated and enhanced from images with strong flame interference. Flame emission is suppressed by near-infrared feature band selection, and visible light motion information is fused for initial localization. Then, frequency domain low-pass filtering is employed to effectively remove dynamic flame texture based on the spectral differences between the flame and cable contours. Finally, morphological post-processing optimizes contour integrity, outputting a high-quality binary cable mask sequence for subsequent analysis. The process is as follows:

[0043] For the acquired visible light image sequences, a dynamic background modeling method based on Gaussian mixture models was used to extract the foreground motion region (including flames and cables). For the near-infrared image sequences, taking advantage of the weak flame radiation, adaptive thresholding and morphological closing operations were used to initially extract the cable region.

[0044] Specifically, the near-infrared narrowband filter with a specific center wavelength is preferably 1.4 μm or 1.9 μm, with a bandwidth of ±20 nm. This band is located in the strong absorption valley of the radiation spectrum of water vapor (H2O) and hydroxyl (OH) in the flame, which can suppress the flame brightness to the greatest extent.

[0045] Adaptive thresholding segmentation employs a Gaussian-Bernson algorithm based on local neighborhoods: for each pixel in the image, the maximum, minimum, and Gaussian-weighted average values ​​within its 15×15 pixel neighborhood are calculated, and the threshold value for that pixel is determined accordingly. ,in In image pixel coordinates The adaptive threshold calculated at that point, These are the maximum and minimum pixel grayscale values ​​within a neighborhood (e.g., 15×15 pixels) centered on the center. The average pixel grayscale value of a neighborhood centered on the center, after Gaussian weighting; , The harmonic coefficients are empirically set to 0.7 and 0.3, respectively. Then, binarization segmentation is performed based on these thresholds.

[0046] By performing a logical AND operation between the moving area in the visible light foreground and the cable area extracted from near-infrared light, the cable portion obscured by the flame is initially separated.

[0047] Furthermore, a two-dimensional discrete Fourier transform is performed on the initially extracted cable region image to analyze its spectrum. Since flame flickering has high-frequency characteristics, while cable contour changes are relatively low-frequency, a second-order two-dimensional Butterworth low-pass filter is designed for frequency domain filtering. Its transfer function is... ,in For a two-dimensional Butterworth low-pass filter at a frequency domain point The transfer function value at that point, It is a frequency point Distance to the center point of the spectrum Here, n=2 represents the cutoff frequency, and n is the order. Cutoff frequency The empirical formula is dynamically set based on the actual physical width of the cable in the image and its pixel ratio. ,in The width of the image (in pixels) is The pixel width of the cable in the image; ensuring that high-frequency textures of flame flickering (typically higher than the dominant frequency of the cable outline) are filtered out while preserving the cable outline information.

[0048] The morphological optimization closure of the cable profile after initial separation is performed as follows:

[0049] The filtered image Perform inverse Fourier transform to obtain the spatial domain enhanced image. Subsequently, regarding Binarization was performed using Otsu's global thresholding method to obtain the initial binary contour map. ;

[0050] To further optimize the integrity of the contour for subsequent accurate analysis, Perform two-level morphological optimization:

[0051] Filling small-area holes: Calculation Area of ​​all background connected regions Define area threshold (like =50 pixels). If a background connected region is completely surrounded by the foreground and its area is 50 pixels. < If it is identified as a noise hole, its pixel value is set from 0 (background) to 1 (foreground), resulting in an intermediate binary image. This operation can eliminate internal gaps caused by localized interference;

[0052] Edge smoothing and bridging of minor breaks: Use a circular structuring element S with radius r (e.g., r = 2 pixels) to smooth edges and bridge minor breaks. Perform morphological closing operations. Closing operations are defined as dilation followed by erosion: ;in, This indicates an expansion operation, which can bridge small fractures with a width less than 2r. This indicates an erosion operation that smooths out edges that have become rough due to expansion, restoring the outline to near its original size. The final optimized binary mask image of the cable region. The image is an intermediate binary image after hole filling, where S is a circular structuring element whose size (e.g., radius r = 2 pixels) determines the range of dilation and erosion operations; the final result is... That is, a complete and smooth cable area mask; Output as a sequence of enhanced, clear outline images of the cable.

[0053] Step 3: 3D Deformation Measurement and Thermal Correction – High-precision quantification of the cable's true 3D deformation from a 2D image sequence. First, a reliable 3D surface is reconstructed by combining stereo matching of speckle features and point cloud filtering based on confidence metrics. Then, an innovative temperature-expansion physical model is established and applied. Using infrared temperature field data, the thermal expansion component of the material is accurately subtracted from the total observed displacement, thereby decoupling and outputting true 3D deformation parameters that only reflect ablation and mechanical damage. The process is as follows:

[0054] Stereo matching was performed on the binocular enhanced image sequence obtained in step two, and the three-dimensional point cloud data of speckle feature points on the cable surface was calculated using calibration parameters. Based on the temperature field distribution data of the cable surface measured by the infrared thermal imager, and combined with the known thermal expansion coefficients of each layer of cable materials (such as conductor, insulation, and sheath), a temperature-expansion mapping model was established.

[0055] Specifically, the stereo matching uses a semi-global block matching algorithm and adds a zero-mean normalized cross-correlation constraint for pre-sprayed speckle to the matching cost function to improve the matching accuracy and robustness under high-temperature image blurring.

[0056] To further improve the reliability of 3D point cloud data, a confidence assessment and filtering mechanism is introduced for the stereo matching results. For each 3D feature point generated by the matching... The matching quality is evaluated jointly by the matching cost function value and the zero-mean normalized cross-correlation value of the corresponding small regions in the left and right images. A comprehensive confidence metric is defined. :

[0057]

[0058] in, Let i be the comprehensive matching confidence score of the i-th 3D feature point. This represents the aggregate cost of the matching point in the semi-global matching. and These are the maximum and minimum cost values ​​among all matching points in the current frame, respectively. The zero-mean normalized cross-correlation value of the image region corresponding to this point is in the range of [-1, 1], and the closer it is to 1, the higher the similarity. This is the weighting coefficient, with an empirical value of 0.4, used to balance the two metrics;

[0059] Set a confidence threshold (For example =0.65). The initial 3D point cloud was calculated. Then, iterate through all points, if a certain point confidence level If a point is found to be a low-confidence match, it is likely due to image blur, occlusion, or incorrect matching and should be removed. This process yields a high-confidence 3D point cloud after noise removal, which is used for subsequent temperature mapping and deformation calculations.

[0060] The temperature-expansion mapping model is specifically expressed as: for a certain point on the cable surface, the three-dimensional coordinate offset caused by thermal expansion. ,satisfy:

[0061]

[0062] in, , , The displacement component of a point on the cable surface on the X, Y, and Z coordinate axes is purely caused by thermal expansion. For this point at the reference room temperature Three-dimensional coordinates at (25℃); The infrared temperature is mapped to this point through joint calibration; To determine the appropriate value based on the cable layer material at that point (e.g., if the insulation layer is cross-linked polyethylene), take... The linear expansion coefficient components determined by axial / radial anisotropy can be simplified to the following for isotropic materials when axial constraints are neglected: .

[0063] For each moment of the calculated 3D point cloud, based on the local temperature corresponding to its spatial location, the model is used to calculate the 3D coordinate offset caused purely by thermal expansion, which is then subtracted from the total measured displacement to correct for the thermal expansion effect, thus obtaining a 3D point cloud of "true deformation" caused only by ablation and mechanical deformation. Based on the corrected 3D point cloud, 3D deformation parameters such as the diameter change rate, axial bending deflection, and cross-sectional ellipticity at key locations of the cable (such as the middle of the suspension point or near the flame) are calculated to form a deformation time series curve.

[0064] Step 4: Multimodal data fusion and failure determination. A unified spatiotemporal benchmark is constructed, synchronously fusing five-dimensional data streams of deformation, ablation, structural continuity, temperature, and circuit status. A time-varying weighted fusion algorithm reflecting the refractory evolution law is designed to calculate comprehensive characteristic indicators. Finally, based on multiple criteria such as abrupt deformation changes, structural fracture, or peak values ​​of fusion indicators, the failure critical point and mode are automatically and accurately determined, and a digital test report is generated, achieving objectivity and automation in the determination process. The process is as follows:

[0065] Construct a time-based multidimensional data pool and synchronously align the following time-series data:

[0066] 1) The key three-dimensional deformation parameters of the cable calculated in step three;

[0067] 2) The proportion of cable surface ablation area calculated by image segmentation in step two using the visible light image sequence;

[0068] 3) Cable structure continuity index in near-infrared image sequences; calculated through the following steps: First, the Zhang-Suen parallel iterative thinning algorithm is applied to the enhanced near-infrared binarized image to extract the cable skeleton with a single pixel width; then, 8-neighborhood tracking is performed along the skeleton pixels. When no next skeleton pixel is found after tracking for more than 5 consecutive pixels, it is determined as a break point; finally, the cable structure continuity index CI is defined as: CI = (total number of skeleton pixels - number of break points × compensation coefficient 10) / total number of skeleton pixels × 100%. The lower the index value, the more severe the cable structure breakage or damage.

[0069] 4) The highest temperature on the unexposed side as measured by an infrared thermal imager;

[0070] 5) Circuit on / off status monitoring for power supply. Design a weighted fusion algorithm to assign different data weights to different stages of testing (e.g., focusing on deformation and temperature in the early stages, and on structural continuity and circuit status in the later stages). Calculate the rate of change of the fused feature vector. The weighted fusion algorithm specifically involves constructing a time-varying weight vector. These correspond to deformation parameters, ablation area, continuity index, back-fire surface temperature, and circuit status, respectively. Among them:

[0071] High initial weight;

[0072] The weight gradually increases in the later stages;

[0073] ;

[0074] and Take a fixed value of 0.1. The sum of the weights is 1.

[0075] Fusion eigenvalues All of these are normalized data. For at any time The calculated scalar fusion eigenvalues For at any time The time-varying weight vector, These are the normalized three-dimensional deformation parameters. This represents the normalized ablation area ratio. The normalized structural continuity index. This represents the normalized highest temperature on the unexposed side. This indicates the on / off state of the circuit; typically, 1 represents on and 0 represents off.

[0076] The judgment thresholds are set based on statistical analysis of a large amount of benchmark fire resistance test data: the threshold for the rate of change of diameter expansion in condition A is set at 5% / second; the threshold for the continuity index in condition B is set at 40%; and the sharp peak value of the rate of change of the fused eigenvector in condition C is defined as... Two consecutive frames exceeded 0.15. This represents the change in the fused feature value over a continuous time interval.

[0077] The cable is automatically determined to have reached the fire resistance failure critical point when any of the following conditions are met:

[0078] A. The rate of change of three-dimensional deformation parameters (such as diameter expansion rate) exceeds a set threshold per unit time and continues to deteriorate;

[0079] B. A sudden drop in the near-infrared structural continuity index below the threshold indicates that the cable structure has broken or a breakdown channel has formed;

[0080] C. A sharp peak appears in the rate of change of the fused eigenvector, accompanied by a circuit break. The system records this moment as the failure time and outputs a complete report on deformation, ablation, and temperature evolution before failure, as well as a failure mode analysis.

[0081] The cable fire resistance testing system modules based on machine vision and image processing are as follows:

[0082] The multispectral synchronous acquisition and high-precision calibration module is used to drive and synchronously control the visible light and near-infrared binocular cameras and infrared thermal imagers to ensure the time alignment of multi-source data. At the same time, it performs geometric calibration and online thermal drift compensation of the binocular system before and during testing. The camera intrinsic parameters are corrected in real time through the established temperature-parameter drift model, providing an accurate and stable spatial geometric reference for all subsequent image measurements and suppressing systematic errors introduced by high-temperature environments from the source.

[0083] The anti-interference cable image enhancement and segmentation module uses a specific near-infrared band to suppress flame emission and fuses visible light motion information to initially locate the cable. Then, it separates the dynamic flame texture from the static cable outline through frequency domain filtering (such as Butterworth low-pass filtering). Finally, it performs morphological optimization (hole filling, closing operation) on the preliminary segmentation results and outputs a complete and smooth sequence of binary cable mask images, providing high-quality input for 3D reconstruction and structural analysis.

[0084] The 3D thermal deformation measurement and physical correction module, based on stereo matching and confidence filtering, reconstructs a highly reliable 3D point cloud of the cable surface from binocular enhanced images. Subsequently, it fuses infrared temperature field data and establishes a temperature-displacement physical model based on the material's thermal expansion coefficient. It then subtracts the component caused purely by thermal expansion from the total observed displacement in real time, thereby outputting 3D deformation parameters (such as diameter change rate and deflection) that reflect the actual ablation and mechanical deformation.

[0085] The multi-source information fusion and intelligent failure determination module is used to construct a unified time axis and synchronously manage multi-dimensional time-series data streams from deformation, ablation area, structural continuity, back surface temperature, and circuit status. Through a designed time-varying weighted fusion algorithm, it calculates comprehensive characteristic indicators and their rate of change. Finally, based on multiple criteria such as pre-set deformation abrupt change, structural fracture, or peak value of fusion indicators, it automatically and accurately determines the critical moment and mode of cable fire resistance failure and generates a complete digital test report.

[0086] The above formulas are all dimensionless calculations. Dimensionless calculations can be performed using various methods such as standardization, which will not be elaborated here. The formulas are derived from software simulations based on a large amount of collected data, and the preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0087] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, ATA hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0088] It should be understood that, in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0089] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0090] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0091] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0092] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0093] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable ATA hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0094] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A cable fire resistance testing method based on machine vision and image processing, characterized in that, Includes the following steps: B1. System Construction and Thermal Stability Calibration: A synchronous acquisition system including a visible-near-infrared binocular vision system, an infrared thermal imager, and a circuit monitoring device was built. The binocular system was calibrated, and online thermal stability compensation was performed based on the temperature-parameter drift model. B2. Multispectral Image Processing and Target Extraction: Acquire visible light and near-infrared images, extract the cable target contour and generate a binary mask sequence through flame suppression, frequency domain filtering and morphological processing; B3. Three-dimensional deformation measurement and thermal correction: Based on stereo matching of binocular images, the three-dimensional point cloud of the cable surface is reconstructed. Combined with infrared temperature data and the material's thermal expansion coefficient, the thermal expansion component is subtracted to obtain the true three-dimensional deformation parameters. B4. Multimodal data fusion and failure determination: Simultaneously fuse deformation, ablation area, structural continuity, temperature and circuit status data, calculate comprehensive characteristic indicators through time-varying weighted fusion algorithm, and automatically determine the critical point and mode of fire resistance failure based on multiple criteria.

2. The cable fire resistance testing method based on machine vision and image processing according to claim 1, characterized in that, The online thermal stability compensation in B1 includes: Before testing, images of the calibration plate at different temperatures were acquired, the rate of change of key calibration parameters with temperature was calculated, and the camera intrinsic parameters were dynamically corrected based on the real-time monitored lens ambient temperature during the test. Key calibration parameters include principal point coordinates, focal length, and lens distortion coefficient.

3. The cable fire resistance testing method based on machine vision and image processing according to claim 1, characterized in that, In B2: Near-infrared narrowband filters of a specific wavelength band are used to suppress flame radiation, and a frequency domain low-pass filter is used to separate the high-frequency texture of flame jumping from the low-frequency information of cable outline.

4. The cable fire resistance testing method based on machine vision and image processing according to claim 1, characterized in that, In B3: The stereo matching algorithm employs a semi-global block matching algorithm and introduces a comprehensive confidence evaluation mechanism based on matching cost and normalized cross-correlation to filter low-confidence 3D points.

5. The cable fire resistance testing method based on machine vision and image processing according to claim 1, characterized in that, In B3: The temperature-expansion mapping model calculates the coordinate offset caused by thermal expansion based on the thermal expansion coefficients of each layer of cable material and infrared temperature measurement data, and deducts it from the total displacement.

6. The cable fire resistance testing method based on machine vision and image processing according to claim 1, characterized in that, In B4: The structural continuity index is calculated by extracting the cable skeleton and detecting break points; The fusion feature index uses a time-varying weight vector to weight and fuse data from different stages.

7. A system applied to the cable fire resistance testing method based on machine vision and image processing as described in any one of claims 1-6, characterized in that, Includes the following modules: The multispectral synchronous acquisition and high-precision calibration module is used to synchronously control the visible-near-infrared binocular camera and the infrared thermal imager, and to perform online thermal drift compensation. An anti-interference cable image enhancement and segmentation module is used to suppress flame interference, extract cable contours, and generate a binary mask sequence. The 3D thermal deformation measurement and physical correction module is used to reconstruct the 3D point cloud of the cable and output the true deformation parameters by subtracting the thermal expansion component based on the temperature data. The multi-source information fusion and intelligent failure determination module is used to fuse multi-dimensional data streams and automatically determine fire resistance failure through time-varying weighted fusion and multiple criteria.

Images