Method for detecting damage of fiber rope in ultra-deep vertical shaft hoisting and safety evaluation

Abstract

A method for damage detection and safety assessment of fiber ropes in ultra-deep vertical shaft hoisting involves simultaneously acquiring terahertz image sequences of internal fiber rope damage and visual image sequences of surface damage using a multi-view camera array monitoring unit located on both sides of the U-shaped detection groove of the hoisting fiber rope, including a terahertz source transmitter, a terahertz imager, and a multi-view camera array at the sheave. Both types of detection data are transmitted to a host computer in real time. Next, the image processing and analysis unit in the host computer performs automatic target identification and preprocessing on the fiber rope detection data. Then, a spatiotemporal synchronous registration method based on SURF feature extraction and RANSAC algorithm optimization is used to obtain precise matching features and spatial mapping relationships of internal and external fiber rope damage. Finally, damage feature extraction and multimodal feature fusion are performed on the spatiotemporally registered multimodal detection data to achieve intelligent identification and precise location of fiber rope damage. Combined with damage quantification indicators, this enables safety assessment and graded early warning of fiber rope operation.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology, specifically to a method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes based on the fusion of terahertz and machine vision. This method enables simultaneous nondestructive testing, intelligent identification, and safety assessment of internal and external damage to ultra-deep vertical shaft hoisting fiber ropes. Background Technology

[0002] As my country's shallow mineral resources gradually deplete, mineral development is increasingly extending to deeper areas, making deep mining a crucial development strategy. Ultra-deep vertical shaft hoisting systems play a vital role in lifting and lowering materials and personnel, serving as the "choke point" connecting the surface and underground. However, with ultra-deep vertical shafts exceeding 2000m in depth, the weight of the hoisting wire ropes has increased significantly. This not only reduces the effective lifting load ratio of the hoisting system but also fails to meet the safety factor requirements stipulated in relevant regulations, becoming a core bottleneck restricting the construction and production of ultra-deep vertical shafts. Fiberglass ropes, with their advantages of light weight and high breaking strength, can effectively reduce the self-weight load of the hoisting system and increase its effective load-bearing capacity, making them an inevitable trend to replace wire ropes in ultra-deep vertical shaft hoisting. However, factors such as the time-varying properties of hoisting fiber ropes, longitudinal torsional coupling characteristics, alternating loads, and external excitations can cause complex impacts and vibrations. Coupled with the effects of harsh environments such as high temperature, high humidity, and corrosive media, these factors can further exacerbate damage to the fiber ropes, including internal and external wear, delamination, and fiber breakage, and may even lead to fracture accidents, severely limiting their service life and load-bearing reliability. Therefore, this invention proposes a damage detection and safety assessment method for ultra-deep vertical shaft hoisting fiber ropes based on the fusion of terahertz and machine vision. This method has significant engineering value and strategic importance for the safe and efficient construction of future ultra-deep vertical shafts and for ensuring the long-term stable development of deep mineral resources.

[0003] Existing hoisting rope damage monitoring technologies include: Patent No. CN201810432434.7 discloses a dynamic detection method for damage to mining hoisting wire ropes. This method acquires dynamic images of the wire rope using a camera, and combines wire rope texture feature extraction methods and BP neural networks to assess surface damage. However, this method can only detect surface damage of wire ropes or fiber ropes and cannot identify internal damage. Patent No. CN202410241593.4 discloses a multi-condition damage detection method for marine winch fiber ropes. This method conducts offline damage detection on fiber ropes under tensile-bending fatigue-friction wear coupled conditions on a self-made fiber rope multi-condition service simulation testing machine, realizing fiber rope load-bearing characteristic analysis and safety assessment. However, this method cannot achieve real-time detection of internal and external damage during the operation of hoisting fiber ropes. Currently, there is no method suitable for real-time detection and safety assessment of internal and external damage to ultra-deep vertical shaft hoisting fiber ropes. Summary of the Invention

[0004] This invention aims to address the problems that fiber ropes may face in future ultra-deep vertical shaft hoisting projects, such as low accuracy of single-modal detection, inability to simultaneously monitor internal and external damage, insufficient damage location accuracy, and imperfect safety assessment system. It provides a method for damage detection and safety assessment of fiber ropes in ultra-deep vertical shaft hoisting based on the fusion of terahertz and machine vision.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes. This method utilizes a terahertz source transmitter, a terahertz imager, and a multi-view camera array monitoring unit located on both sides of the U-shaped groove for detecting the hoisting fiber rope. These elements, along with a terahertz source transmitter, a terahertz imager, and a headwheel, simultaneously acquire terahertz image sequences of internal damage and visual image sequences of surface damage within the hoisting fiber rope. A spatiotemporal synchronous registration method optimized by SURF feature extraction and RANSAC algorithm is used to achieve accurate matching of internal and external damage data for the fiber rope. Finally, damage feature extraction and Laplacian pyramid multimodal feature fusion are performed to complete intelligent identification, precise positioning, safety assessment, and graded early warning of fiber rope damage.

[0006] A method for damage detection and safety assessment of fiber ropes used in ultra-deep vertical shaft hoisting includes the following steps:

[0007] Step 1: A collaborative detection system is formed by arranging a multi-view camera array monitoring unit on both sides of the fiber rope detection U-shaped groove (between the rope exit end of the hoist drum and the sheave) with a terahertz source transmitter, a terahertz imager, and the sheave. The system simultaneously acquires terahertz image sequences of internal damage and visual image sequences of surface damage in the fiber rope and transmits the two types of detection data to the host computer in real time.

[0008] Step 2: The image processing and analysis unit in the host computer performs automatic target identification and preprocessing on the fiber rope detection data. Then, the spatiotemporal synchronous registration method based on SURF feature extraction and RANSAC algorithm optimization is used to obtain the accurate matching features and spatial mapping relationship of internal and external damage of the fiber rope.

[0009] Step 3: Extract damage features and fuse multimodal features from the spatiotemporally registered multimodal detection data obtained in Step 2 to complete intelligent identification and precise positioning of fiber rope damage. Combine damage quantification indicators to achieve fiber rope operation safety assessment and graded early warning.

[0010] Preferably, the specific process of automatically identifying and preprocessing the fiber rope detection data and achieving spatiotemporal synchronous registration in step 2 is as follows:

[0011] Step 2.1: Perform automatic identification and dynamic tracking of the target fiber rope ROI on the terahertz image and machine vision image, extract the fiber rope detection sub-image in the ROI region, and convert the detection sub-image into a grayscale image to automatically shield background interference, reduce data dimensionality, and improve the algorithm's ability to resist rope swaying and its operating efficiency.

[0012] Step 2.2: Perform median filtering and adaptive gamma correction on the grayscale image to enhance the damaged edges, local textures and details of the fiber rope in the grayscale image, and obtain a feature-enhanced grayscale image of the fiber rope.

[0013] Step 2.3: Perform multi-scale SURF feature point extraction on the enhanced terahertz image and visual image within the ROI region to construct a feature descriptor of uniform dimension and complete the initial matching of fiber rope stability features; the multi-scale SURF feature point extraction is based on Gaussian filtering and Hessian matrix, and the formula for calculating the Hessian matrix is:

[0014] ;

[0015] In the formula, H(x, y, σ) is the Hessian matrix; , and denoted as the second-order partial derivatives in the x, xy, and yy directions, respectively.

[0016] Step 2.4: The RANSAC algorithm is used to remove outliers from the initial feature matching pairs to optimize feature matching accuracy and eliminate mismatches caused by rope swaying, lighting fluctuations, and environmental noise. The formula for calculating the number of iterations required by the algorithm is as follows: ;

[0017] In the formula, k is the required number of iterations; p is the probability that the algorithm successfully samples the "pure interior point set"; e is the proportion of mismatches in the initial feature matching pairs, based on the statistical characteristics of fiber rope bimodal images; and n is the minimum number of feature point pairs required to fit the homography matrix.

[0018] Step 2.5: Perform threshold segmentation and morphological opening and closing operations on the optimized matching image from Step 2.4 to achieve accurate separation of the rope region from the background and smooth the contour of the damaged area.

[0019] Step 2.6: Based on the image processed in Step 2.5, and combined with the optimized precise feature matching pairs in Step 2.4, construct a homography matrix to realize the pixel coordinate space mapping between terahertz internal damage data and visual surface damage data, and complete the precise spatiotemporal synchronous registration of the two types of data in the axial, circumferential and radial dimensions.

[0020] Step 2.3, multi-scale SURF feature point extraction, is based on Gaussian filtering and the Hessian matrix. The formula for calculating the Hessian matrix is: ;

[0021] In the formula, The Hessian matrix; , and denoted as the second-order partial derivatives in the x, xy, and yy directions, respectively.

[0022] In step 2.4, when using the RANSAC algorithm to remove outliers, the formula for calculating the required number of iterations is as follows: ;

[0023] In the formula, k is the required number of iterations; p is the probability that the algorithm successfully samples the "pure interior point set"; e is the proportion of mismatches in the initial feature matching pairs, based on the statistical characteristics of fiber rope bimodal images; and n is the minimum number of feature point pairs required to fit the homography matrix.

[0024] Preferably, step 3, which involves extracting damage features and fusing multimodal features from the spatiotemporally registered multimodal detection data, and completing damage identification, localization, and safety assessment, includes the following sub-steps:

[0025] Step 3.1: Damage features are extracted modally from the spatiotemporally registered terahertz image and the visual image. The terahertz image extracts internal damage features such as internal layering of the fiber rope, fiber breakage, and matrix voids. The visual image extracts surface damage features such as broken fibers, wear, and corrosion. Laplacian pyramid multi-scale feature fusion is applied to the dual-modal images. The fusion calculation formula is as follows: And satisfy ;

[0026] In the formula, The fused Laplacian feature map of the i-th layer and These are the Laplacian pyramid features of the terahertz image and the visual image at the i-th layer, respectively. and These are the fusion weights for the terahertz mode and the visual mode, respectively.

[0027] Step 3.2: Normalize the damage features extracted from the different modes, and use an attention-weighted feature fusion algorithm based on feature importance assessment to assign adaptive weights to the damage features of different modes, and construct a multimodal fusion feature vector of internal and external damage of the fiber rope.

[0028] Step 3.3: Input the multimodal fusion feature vector into the pre-trained and optimized ResNet-50 bimodal damage classification network to complete the intelligent identification of fiber rope damage type. At the same time, combine the location information of ROI dynamic tracking with the spatial mapping relationship established in step 2 to accurately locate the axial position, circumferential range and radial depth of the damage.

[0029] Step 3.4: Based on the identified and located damage data, establish a quantitative index system for damage severity. Through core parameters such as damage size, defect percentage, number of broken fibers, and wear rate, complete the quantitative grading of fiber rope damage. Specifically, the Gray Moment algorithm is used for damage quantification, with the core being the second-order center gray moment formula and the damage severity quantification formula: ;

[0030] In the formula, m 22 The second-order central gray-level moment of the damaged region characterizes the non-uniformity of gray-level distribution in the damaged region; and The average pixel coordinates of the damaged area. is the pixel grayscale value, and S is the total number of pixels in the damaged area; It is the 0th order grayscale moment (total number of pixels in the damaged area). is the normalized grayscale moment; D is the damage degree quantization value (range 0-1, the larger the value, the more severe the damage), and k is the quantization coefficient.

[0031] Step 3.5: Combining the quantitative grading results of fiber rope damage with operating parameters such as the operating load, lifting speed, and running time of the lifting system, construct a fiber rope safety status assessment model. Through this model, first calculate the comprehensive risk value of damage, then derive the safety and reliability coefficient of the fiber rope, comprehensively determine the current operating safety status of the fiber rope, and trigger the corresponding level of early warning signal according to the damage level, and output a fiber rope non-destructive testing and safety assessment report.

[0032] Based on the quantitative grading results of fiber rope damage and operating parameters such as hoisting system load, hoisting speed, and operating time, a fiber rope safety status assessment model is constructed. This model first calculates the comprehensive damage risk value, and then derives the fiber rope safety reliability coefficient. The formula for calculating the comprehensive damage risk value is as follows: ;

[0033] In the formula, R is the comprehensive risk value of damage, ranging from 0 to 1. The larger the value, the higher the risk of fiber rope operation; ω is the damage degree weighting coefficient, ranging from 0.6 to 0.8, used to characterize the proportion of the impact of the severity of damage on the safety status; D is the quantitative value of damage degree obtained in step 3.4; T is the damage type risk coefficient, which is assigned according to the degree of damage hazard. Severe damage such as rope core breakage and internal delamination is assigned 0.8-1.0, broken wires are assigned 0.3-0.5, and minor damage such as wear is assigned 0.1-0.2.

[0034] In step 3.1, the spatiotemporally registered dual-modal images are fused using Laplacian pyramid multi-scale features. The fusion calculation formula is as follows: And satisfy ;

[0035] In the formula, The fused Laplacian feature map of the i-th layer and These are the Laplacian pyramid features of the terahertz image and the visual image at the i-th layer, respectively. and These are the fusion weights for the terahertz mode and the visual mode, respectively.

[0036] In step 3.3, the loss function of the ResNet-50 bimodal damage classification network is the cross-entropy loss function, calculated as follows: ;

[0037] In the formula, The loss is the classification loss; N is the batch size. Let i be the true label of the i-th sample; denoted as the predicted probability of the i-th sample; C represents the number of damage categories, including broken fibers, abrasion, delamination, and core breakage. The network minimizes the classification loss through backpropagation, thereby improving the accuracy of fiber rope damage type identification.

[0038] Step 3.4 uses the Gray Moment gray-moment damage quantification and grading algorithm to quantify the located damage area. The core calculation formulas include the second-order center gray-moment formula and the normalized gray-moment formula: ;

[0039] In the formula, m 22 The second-order central gray-level moment of the damaged region characterizes the non-uniformity of gray-level distribution in the damaged region; and The average pixel coordinates of the damaged area. is the pixel grayscale value, and S is the total number of pixels in the damaged area; It is the 0th order grayscale moment (total number of pixels in the damaged area). is the normalized grayscale moment; D is the damage degree quantization value (range 0-1, the larger the value, the more severe the damage), and k is the quantization coefficient.

[0040] The formula for calculating the comprehensive risk value of fiber rope damage in step 3.5, combining the damage type weights, is as follows: ;

[0041] In the formula, R is the comprehensive risk value of damage, ranging from 0 to 1. The larger the value, the higher the risk of fiber rope operation; ω is the damage degree weighting coefficient, ranging from 0.6 to 0.8, used to characterize the proportion of the impact of the severity of damage on the safety status; D is the quantitative value of damage degree obtained in step 3.4; T is the damage type risk coefficient, which is assigned according to the degree of damage hazard. Severe damage such as rope core breakage and internal delamination is assigned 0.8-1.0, broken wires are assigned 0.3-0.5, and minor damage such as wear is assigned 0.1-0.2.

[0042] Preferably, the area array terahertz source transmitter operates in the 0.1-1THz frequency band and has a spatial resolution better than 0.5mm; the terahertz imager is an area array imaging detector with a maximum sampling frame rate of 60Hz; the multi-view camera array monitoring unit includes three high-speed industrial cameras, each with a maximum shooting frame rate of 120Hz and a resolution of 3840×2160. The three high-speed industrial cameras are evenly fixed on the ring camera bracket by clamps, and the camera field of view completely covers the circumferential surface of the lifting fiber rope, realizing the acquisition of visual images of the fiber rope surface without blind spots; the ring camera bracket is fixed on the multi-view camera array monitoring unit bracket at the sheave, maintaining a preset safe detection distance from the lifting fiber rope to avoid contact with the rope and causing operational interference.

[0043] Preferably, the multi-camera array monitoring unit is simplified to two high-speed industrial cameras. It adopts a single-scale SURF feature point extraction combined with a RANSAC algorithm with a fixed number of iterations, a simple weighted fusion algorithm, and a lightweight CNN-like network. It only realizes the axial and circumferential positioning of fiber rope damage and rapid determination of safety status, omitting the radial depth fine calibration, micro-defect detection, and complex parameter quantization steps.

[0044] Compared with the prior art, the beneficial effects of the present invention are:

[0045] (1) A real-time detection method for internal and external damage of hoisting fiber rope in ultra-deep vertical shaft based on terahertz and machine vision detection technology was proposed. A spatiotemporal synchronous registration method coupled with SURF feature extraction and RANSAC algorithm optimization was proposed. This effectively solved the problems that a single detection method cannot detect internal and external damage at the same time, the registration deviation is large under complex working conditions, and the damage location is inaccurate. This improved the stability and positioning accuracy of internal and external damage detection of hoisting fiber rope.

[0046] (2) A damage quantification and hierarchical early warning system for ultra-deep vertical shaft hoisting fiber ropes based on a multi-modal data fusion strategy (coupled with Laplacian pyramid multi-scale feature fusion and ResNet-50 network damage identification) and Gray Moment method was proposed. This system effectively solved the problems of low reliability of fiber rope damage detection by single-modal data and inaccurate damage assessment. It also achieved accurate identification and comprehensive assessment of damage type, three-dimensional damage location and damage deterioration degree of ultra-deep vertical shaft hoisting fiber ropes. Attached Figure Description

[0047] Figure 1 Schematic diagram of a simulated hoisting fiber rope damage detection system for ultra-deep vertical shafts;

[0048] Figure 2 This is a schematic diagram of the drive and execution unit structure of a simulated hoist in an ultra-deep vertical shaft.

[0049] Figure 3 This is a schematic diagram of the structure of a multi-view camera array monitoring unit;

[0050] Figure 4 Flowchart of the data processing method for fiber rope hoisting in ultra-deep vertical shafts.

[0051] The components include: 1. Hoist base; 2. Hoist; 2-1. Control cabinet; 2-2. Drive motor; 2-3. Drum; 2-4. Double-folded rope groove; 2-5. Drum support; 3. Hoisting fiber rope; 4. Fiber rope detection U-shaped groove; 5. Area array terahertz source transmitter support; 6. Area array terahertz source transmitter; 7. Terahertz imager support; 8. Terahertz imager; 9. Sheave support; 10. Sheave; 11. Multi-view camera array monitoring unit; 11-1. High-speed industrial camera; 11-2. Fixture; 11-3. Ring camera support; 11-4. Signal line; 12. Multi-view camera array monitoring unit support; 13. Tank passage; 14. Frame; 15. Wedge rope ring; 16. Tank ear; 17. Hoisting load; 18. Host computer. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0053] Example 1

[0054] Depend on Figure 1It is known that the ultra-deep vertical shaft simulated hoisting fiber rope damage detection system includes: hoist base 1, hoist 2, hoisting fiber rope 3, fiber rope detection U-shaped groove 4, area array terahertz source transmitter bracket 5, area array terahertz source transmitter 6, terahertz imager bracket 7, terahertz imager 8, sheave bracket 9, sheave 10, multi-view camera array monitoring unit 11, multi-view camera array monitoring unit bracket 12, guideway 13, frame 14, wedge rope ring 15, sheave ear 16, hoisting load 17, host computer 18.

[0055] Furthermore, the hoist 2, as the core actuator of the simulated hoisting system, has its drum connected to the drive unit, converting the rotational torque output by the drive motor into the linear motion of the lifting fiber rope 3, providing traction power for the lifting and lowering of the load 17. The fiber rope detection U-shaped groove 4 is a terahertz detection area located between the rope exit end of the hoist 2 drum and the sheave 10. The lifting fiber rope 3 passes through the groove and operates without contact with the groove body, avoiding running resistance or wear caused by contact between the groove body and the rope body, and ensuring the interference-free coordinated operation of the detection system and the hoisting system.

[0056] The lifting fiber rope 3 is a braided lifting rope made of ultra-high molecular weight polyethylene. One end passes around the sheave 10 and is rigidly connected to the lifting load 17 through the wedge-shaped rope ring 15. The other end is wound in an orderly manner and fixed on the drum of the hoist 2 according to the preset number of winding layers and turns.

[0057] The array terahertz source transmitter 6 is fixed to one side of the lifting fiber rope 3 via the array terahertz source transmitter bracket 5, and its emission optical axis is perpendicular to the radial center of the fiber rope, emitting continuous terahertz waves into the rope body; the terahertz imager 8 is fixed to the opposite side of the lifting fiber rope 3 via the terahertz imager bracket 7, and is coaxially aligned with the array terahertz source transmitter 6 in a transmission manner, acquiring the terahertz wave signal after penetrating the fiber rope, and generating a terahertz image sequence of the internal structure of the fiber rope.

[0058] Furthermore, the terahertz source transmitter 6 operates in the 0.1-1THz frequency band and has a spatial resolution better than 0.5mm, which can meet the requirements for improving the detection accuracy of micro-damage inside the fiber rope 3; the terahertz imager 8 is an array imaging detector with a maximum sampling frame rate of 60Hz, which can realize continuous dynamic imaging during the operation of the fiber rope.

[0059] The sheave 10 is fixedly mounted on the top of the frame 14 via the sheave bracket 9, and is used to support and change the lifting direction of the lifting fiber rope 3. The frame 14 is a metal frame structure, providing a stable and reliable mounting foundation for the sheave 10, terahertz imager, area array terahertz source transmitter, and multi-view camera array monitoring unit, and bearing the operating load of the lifting system.

[0060] The lifting load 17 is a standard counterweight used to simulate the actual load conditions during the hoisting process of an ultra-deep vertical shaft. It is connected to the lower end of the hoisting fiber rope 3 through a wedge-shaped rope loop 15 and moves up and down along the guide rail 13 under the traction of the fiber rope. The guide rail 13 and the lugs 16 are used to provide guidance and constraints for the operation of the lifting load 17, limiting its lateral swing and replicating the real operating conditions of ultra-deep vertical shaft hoisting.

[0061] Depend on Figure 2 As can be seen, the ultra-deep vertical shaft simulation hoist 2 includes a control cabinet 2-1, a drive motor 2-2, a drum 2-3, a double-folded rope groove 2-4, and a drum support 2-5. The drive motor 2-2, as the core power source of the simulation hoisting system, has its output shaft connected to the main shaft of the drum 2-3 via a coupling, converting the rotational torque output by the motor into the linear reciprocating motion of the wound fiber rope 3, providing traction power and speed control for simulating ultra-deep vertical shaft hoisting conditions. The drum support 2-5 provides stable support and installation reference for the drum 2-3, ensuring coaxiality and stability during drum operation. One end of the hoisting fiber rope 3 is wound in an orderly manner within the double-folded rope groove 2-4 of the drum 2-3 according to a preset number of layers and turns, while the other end passes sequentially through the preset detection channels of the terahertz imaging detection unit and the multi-view vision acquisition unit, achieving synchronous acquisition of dual-modal data.

[0062] Depend on Figure 3 It is known that the multi-camera array monitoring unit 11 includes 3 high-speed industrial cameras 11-1. The cameras are evenly fixed on the ring camera bracket 11-3 by the clamp 11-2. The ring camera bracket 11-3 is fixed on the multi-camera array monitoring unit bracket 12 at the sheave, maintaining a preset safe detection distance from the lifting fiber rope 3 to avoid contact with the rope and causing operational interference.

[0063] Depend on Figure 4 The specific steps for damage detection and safety assessment of fiber ropes in ultra-deep vertical shaft hoisting are as follows:

[0064] Step 1: Use a terahertz imager and a multi-view camera array monitoring unit to form a dual-modal collaborative acquisition system to simultaneously acquire terahertz image sequences and multi-view machine vision image sequences of the hoisting fiber rope under simulated hoisting conditions in ultra-deep vertical shafts, and transmit the dual-modal image data to the host computer in real time;

[0065] Step 2: Preprocess, extract features, and remove mismatches from the original dual-modal image sequence in the host computer 18, and finally establish the spatial mapping relationship between internal and external damage data to complete accurate spatiotemporal synchronization registration. The specific steps are as follows:

[0066] Step 2.1: Complete automatic ROI recognition and dynamic tracking in terahertz image sequence and multi-view machine vision image sequence respectively, and take fiber rope detection sub-images in ROI area, convert the detection sub-images into grayscale images, remove background interference areas, focus on fiber rope detection subject, reduce data dimensionality, and improve the running efficiency of subsequent algorithms and the ability to resist rope swaying interference.

[0067] Step 2.2: The dual-modal grayscale image is sequentially subjected to median filtering for noise reduction and adaptive gamma correction for contrast enhancement. This process suppresses image noise, enhances the edges, local textures, and details of damage inside and outside the fiber rope, and yields preprocessed terahertz grayscale and visual grayscale images. The formula for adaptive gamma correction is as follows: ;

[0068] In the formula, The enhanced image pixel values; I in These are the pixel values ​​of the image before preprocessing. These are adaptive correction coefficients; : grayscale value of the pixel neighborhood; k is the contrast adjustment gain coefficient; T: local contrast threshold; local contrast of the pixel. =max (i, j)∈N I in (i,j)-min (i, j)∈N I in (i,j), N is a 5×5 neighborhood of (x,y).

[0069] Step 2.3: Perform multi-scale SURF feature point extraction on the preprocessed terahertz grayscale image and the visual grayscale image to obtain the key feature point set of the dual-modal image. This is specifically implemented based on Gaussian filtering and the Hessian matrix.

[0070] 2.3.1: Multi-scale SURF feature point extraction is performed on the preprocessed terahertz grayscale image and the visual grayscale image to obtain the key feature point set of the dual-modal image;

[0071] 2.3.2: Constructing the Hessian matrix at various scales: ;

[0072] In the formula, H(x, y, σ) is the Hessian matrix; , and denoted as the second-order partial derivatives in the x, xy, and yy directions, respectively.

[0073] 2.3.3: Calculate the approximate determinant of the Hessian matrix and select characteristic points: ;

[0074] In the formula, det(H) is the determinant of the Hessian matrix, used to determine the saliency of feature points.

[0075] Step 2.4: Based on the Euclidean distance matching criterion, perform initial matching on the feature point set of the bimodal image to obtain the initial feature matching pairs of the bimodal image;

[0076] Step 2.5: Based on the core logic of iterative sampling-model fitting-interior point screening, the RANSAC algorithm is used to remove mismatched pairs from the initial feature matching pairs. The specific steps are as follows:

[0077] 2.5.1: Calculate the number of iterations required by the algorithm: ;

[0078] In the formula, K is the required number of iterations; p is the probability that the algorithm successfully samples the "pure interior point set"; e is the proportion of mismatches in the initial feature matching pairs, based on the statistical characteristics of fiber rope bimodal images; and n is the minimum number of feature point pairs required to fit the homography matrix.

[0079] 2.5.2: During the iteration process, each random sample fits a homography matrix H to the feature points, and the interior point determination formula is used: ;

[0080] In the formula, d represents the reprojection error (Euclidean distance) of the feature point pair; These are the pixel coordinates of feature points in a terahertz image; These are the homogeneous coordinates of the corresponding feature points in the visual image;

[0081] 2.5.3: After the iteration, the homography matrix with the largest number of interior points is selected as the optimal model: Based on this model, all inliers are retained as accurate feature matching pairs, while outliers (false matching pairs) are removed.

[0082] In the formula, H opt : Optimal homography matrix (for image registration); H i : The homography matrix fitted in the i-th iteration; N in (H i ): The number of feature point pairs that satisfy the interior point determination in the i-th iteration.

[0083] Step 2.6: Based on accurate feature matching pairs, complete the bimodal image spatial mapping, and use homography matrix to realize the mapping transformation of pixel coordinates to complete the spatiotemporal registration of terahertz and visual images.

[0084] Furthermore, the specific steps of the bimodal image space mapping are as follows:

[0085] 2.6.1: Based on the accurate feature matching pairs obtained in step 2.5.3, the homography matrix H is solved using the least squares method. The core solution formula is: ;

[0086] In the formula, Let be the homogeneous coordinates of the visual image feature points in the pair of matching points; Let be the homogeneous coordinates of the terahertz image feature points in the th pair of matching points; homography matrix. ; L2 is the square of the norm, representing the sum of squares of the reprojection error. The objective is to minimize this error.

[0087] 2.6.2: Substitute the homogeneous coordinates of all pixels in the visual image into the solved homography matrix to complete the mapping to the terahertz image coordinate system: ;

[0088] The mapping result is homogeneously normalized to obtain the pixel coordinates in the terahertz image coordinate system: ;

[0089] In the formula, Let (x′, y′, w) be the original coordinates of any pixel in the visual image; (x′, y′, w) be the homogeneous coordinates after matrix mapping; , ) These are the pixel coordinates mapped to the terahertz image coordinate system.

[0090] 2.6.3: For the non-integer pixel coordinates obtained after mapping, a bilinear interpolation algorithm is used to supplement the pixel gray values ​​to ensure the continuity of the registered image; using the terahertz image as the reference coordinate system, the mapped visual image is aligned with the terahertz image to complete the spatiotemporal registration of the dual-modal image and output the registered dual-modal image pair.

[0091] Step 3: Perform multimodal feature fusion, damage localization and identification, and damage degree quantification and classification on the spatiotemporally registered dual-modal images to complete the fiber rope safety status assessment and graded early warning. The specific steps are as follows:

[0092] Step 3.1: The Laplacian pyramid multi-scale feature fusion is applied to the spatiotemporally registered bimodal image to weight and fuse the terahertz internal damage features with the machine vision surface damage features to generate a unified bimodal fusion feature map.

[0093] Furthermore, the calculation formula for the Laplacian multi-scale feature fusion is as follows: And satisfy ;

[0094] In the formula, The fused Laplacian feature map of the i-th layer and These are the Laplacian pyramid features of the terahertz image and the visual image at the i-th layer, respectively. and These are the fusion weights for the terahertz mode and the visual mode, respectively.

[0095] Step 3.2: Input the dual-modal fused feature map into the ResNet-50 dual-modal damage classification network to complete the intelligent identification of the damage type of the lifting fiber rope. The loss function of this network adopts the cross-entropy loss function, and the calculation formula is as follows:

[0096] ;

[0097] In the formula, The loss is the classification loss; N is the batch size. Let i be the true label of the i-th sample; denoted as the predicted probability of the i-th sample; C represents the number of damage categories, including broken fibers, abrasion, delamination, and core breakage. The network minimizes the classification loss through backpropagation, thereby improving the accuracy of fiber rope damage type identification.

[0098] Step 3.3: Combining the output results of the ResNet-50 bimodal damage classification network, relying on the pixel coordinate-physical coordinate mapping relationship established after bimodal image registration, and the position information of ROI dynamic tracking, the accurate identification and three-dimensional spatial positioning of the damage type of the lifting fiber rope are completed, and the axial position, circumferential angle and radial depth parameters of the damage are finally output.

[0099] Furthermore, the mapping and positioning calculation steps between pixel coordinates and physical coordinates are as follows:

[0100] Step 3.3.1: Establish the basic mapping relationship between pixel coordinates and physical coordinates: ;

[0101] Step 3.3.2: Based on the basic mapping relationship, calculate the axial location of the fiber rope damage (along the fiber rope length direction): ;

[0102] Step 3.3.3: Combining the basic mapping relationship with the grayscale features of the terahertz image, calculate the radial depth of the fiber rope damage: .

[0103] In the formula, The physical coordinates of the damaged area; K represents the pixel coordinates of the damaged area; K is the coordinate scaling factor; B is the coordinate offset. For the axial length of the damage, and These are the physical coordinates of the two ends of the damaged area; Where is the radial depth of the damage, and 'a' is the depth calibration coefficient; The grayscale values ​​of the feature map are fused to represent the damaged area. This represents the grayscale value of the normal area.

[0104] Step 3.4: The Gray Moment (GMT) damage quantification and grading algorithm is used to quantify the located damaged areas and complete the grading of the damage to the lifting fiber rope. The quantification results are directly used for subsequent safety assessments.

[0105] Furthermore, the specific steps for quantifying the degree of damage in the Gray Moment are as follows:

[0106] Step 3.4.1: Calculate the second-order central gray-scale moment of the damaged region: ;

[0107] Step 3.4.2: Normalize the second-order center gray-scale moment (to eliminate the influence of image size): ;

[0108] Step 3.4.3: Calculate the quantization value of the damage degree based on the normalized gray moment to obtain the final quantization result: ;

[0109] In the formula, m 22 The second-order central gray-level moment of the damaged region characterizes the non-uniformity of gray-level distribution in the damaged region; and The average pixel coordinates of the damaged area. is the pixel grayscale value, and S is the total number of pixels in the damaged area; It is the 0th order grayscale moment (i.e., the total number of pixels in the damaged area). is the normalized grayscale moment; D is the damage degree quantization value (range 0-1, the larger the value, the more severe the damage), and k is the quantization coefficient.

[0110] Step 3.5: Based on the damage type, spatial positioning results, and damage degree quantification, a graded threshold judgment method is adopted, combined with the damage type weight, to complete the safety status assessment and graded early warning of the lifting fiber rope, outputting three-level status signals of safety, early warning, and alarm, and simultaneously triggering the corresponding prompt mechanism.

[0111] Furthermore, the specific calculation and judgment steps for the tiered early warning are as follows:

[0112] Step 3.5.1: Calculate the comprehensive risk value of fiber rope damage by combining the damage type weights: ;

[0113] Step 3.5.2: Based on the comprehensive risk value, the graded threshold determination method is used to classify the safety status level of the fiber rope:

[0114]

[0115] In the formula, R is the comprehensive damage risk value, ranging from 0 to 1, with a larger value indicating a higher risk of fiber rope operation; ω is the damage severity weighting coefficient, ranging from 0.6 to 0.8, used to characterize the proportion of the impact of damage severity on the safety status; D is the quantitative value of damage severity obtained in step 3.4, ranging from 0 to 1; T is the damage type risk coefficient, assigned according to the degree of damage hazard, with severe damage such as core breakage and internal delamination taking 0.8-1.0, multiple concentrated broken wires taking 0.6-0.7, and minor surface wear and scratches taking 0.1-0.2; S is the safety status level, with S=1 indicating safety, S=2 indicating warning, and S=3 indicating alarm; D1 and D2 are risk classification thresholds, determined by fiber rope service safety standards and engineering practices.

[0116] Example 2

[0117] This embodiment addresses the application scenario of daily rapid inspection and low-precision screening of hoisting fiber ropes in ultra-deep vertical shafts. Based on the ultra-deep vertical shaft simulated hoisting fiber rope damage detection system based on terahertz and machine vision in Embodiment 1, it retains its overall structure, only making lightweight and simplified optimizations to the equipment acquisition parameters, number of cameras, and algorithm processing flow. The core equipment layout remains the same. Figure 1-3 The corresponding relationships remain consistent and will not be elaborated further. By simplifying the detection process, reducing algorithm complexity, and optimizing equipment acquisition parameters, the detection efficiency is improved while ensuring effective identification of major damages. The method of this invention is further explained below, with the specific technical solution as follows:

[0118] Step 1: Use the terahertz imager 8 and the multi-view camera array monitoring unit 11 to form a dual-modal acquisition system, simultaneously acquire the terahertz image sequence and machine vision image sequence of the lifting fiber rope 3, and transmit the two types of data to the host computer 18 in real time.

[0119] Furthermore, the area array terahertz source transmitter 6 and the terahertz imager 8 are still symmetrically fixed on both sides of the fiber rope detection U-shaped groove 4 and maintain transmission coaxial alignment. The area array terahertz source transmitter 6 operates in the frequency band of 0.1-1THz and has a spatial resolution better than 0.5mm. The terahertz imager 8 is an area array imaging detector with a sampling frame rate set to 60Hz. The multi-view camera array monitoring unit 11 at the sheave 10 is simplified from 3 high-speed industrial cameras to 2. The 2 cameras are evenly fixed on the ring camera bracket by clamps. The highest shooting frame rate of a single camera is 120Hz and the resolution is 3840×2160. The sampling frame rate is reduced to 60Hz. The field of view of the 2 cameras covers the circumferential surface of the lifting fiber rope 3 in a coordinated manner, realizing the rapid acquisition of surface damage visual images without blind spots. All detection data are transmitted to the host computer 18 in real time.

[0120] Step 2: The dual-modal image sequence is processed by the host computer 18 in a lightweight manner, omitting steps such as fine enhancement and complex iterative calculations, to quickly complete target recognition and spatiotemporal registration, only realizing the spatial mapping of fiber rope damage axial and circumferential directions. The specific steps are as follows:

[0121] Step 2.1: Automatically identify the ROI of the target fiber rope in the terahertz image sequence and machine vision image sequence, extract the detection sub-images within the ROI area and convert them into grayscale images, and remove background interference;

[0122] Step 2.2: Denoise the grayscale image by applying a 3×3 window mid-range filter, omitting the fine contrast enhancement step, and directly obtaining a grayscale image with recognizable basic features;

[0123] Step 2.3: Extract single-scale SURF feature points from the grayscale image within the ROI region (only retain the 5×5 window feature calculation), construct a simplified feature descriptor, and complete the initial matching;

[0124] Step 2.4: Use the RANSAC algorithm with a fixed number of iterations (50 times) to remove outliers from the initial feature matching pairs, optimize the matching accuracy, and omit the step of adaptive calculation of the number of iterations;

[0125] Step 2.5: Apply a global fixed threshold segmentation to the optimized matching image to separate the rope region from the background, omitting complex post-processing steps such as morphological opening and hole filling;

[0126] Step 2.6: Based on the segmented images, establish the axial and circumferential spatial mapping relationship between terahertz internal damage data and visual surface damage data, omitting the fine calibration and mapping of radial depth, and quickly complete the simple spatiotemporal synchronization registration of the two types of data.

[0127] Step 3: Perform lightweight feature fusion and recognition on the spatiotemporally registered dual-modal images, omitting steps such as minor defect detection and complex parameter quantization, and only achieving the identification and classification of major damages and rapid determination of safety status. The specific steps are as follows:

[0128] Step 3.1: Extract damage features from the spatiotemporally registered terahertz image and the visual image. The terahertz image focuses on major internal damage such as internal delamination and fiber breakage, while the visual image focuses on major surface damage such as broken filaments and obvious wear. Extraction of minor defect features is omitted.

[0129] Step 3.2: Normalize the extracted damage features, and use a simple weighted fusion algorithm to assign fixed weights (0.5 for terahertz modality and 0.5 for visual modality) to construct a multimodal fusion feature vector, omitting adaptive weight calculation;

[0130] Step 3.3: Input the fused feature vector into the lightweight CNN classification network to complete the intelligent identification of three main damage types: broken wire, wear, and delamination. Combine the ROI location information to locate the axial position and circumferential range of the damage.

[0131] Step 3.4: Based on the damage data after identification and positioning, only two core parameters, damage size and number of broken fibers, are selected. The grading standard calibrated in engineering practice is used to complete the simplified quantitative grading of the degree of fiber rope damage (mild, moderate, severe), omitting the calculation of complex parameters such as defect ratio and wear rate.

[0132] Step 3.5: Based on the damage classification results and referring to the basic operating parameters such as the normal operating load and lifting speed of the hoisting system, determine the current safe operating status of the fiber rope, and output three levels of signals: safety, early warning, and alarm, as well as a simplified detection report (only including the damage type, approximate location, and safety level, omitting detailed data statistics).

[0133] Although embodiments of the invention have been shown and described (see the detailed description above), it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for damage detection and safety assessment of fiber ropes used in ultra-deep vertical shaft hoisting, characterized in that, By deploying a terahertz source transmitter, a terahertz imager, and a multi-view camera array monitoring unit at the sheave on both sides of the U-shaped groove for lifting fiber rope detection, terahertz image sequences of internal damage and visual image sequences of surface damage in the lifting fiber rope are acquired simultaneously. The spatiotemporal synchronous registration method optimized by SURF feature extraction and RANSAC algorithm is used to achieve accurate matching of internal and external damage data of the fiber rope. Then, through damage feature extraction and Laplacian pyramid multimodal feature fusion, intelligent identification, accurate positioning, safety assessment, and graded early warning of fiber rope damage are completed.

2. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 1, characterized in that, Specifically, the following steps are included: Step 1: A collaborative detection system is formed by the array terahertz source emitter and terahertz imager on both sides of the fiber rope detection U-shaped groove between the rope exit end of the hoist drum and the sheave, and the multi-view camera array monitoring unit at the sheave. The system simultaneously collects terahertz image sequences of internal damage and visual image sequences of surface damage in the hoisting fiber rope, and transmits the two types of detection data to the host computer in real time. Step 2: The image processing and analysis unit in the host computer performs automatic target identification and preprocessing on the fiber rope detection data. Then, the spatiotemporal synchronous registration method based on SURF feature extraction and RANSAC algorithm optimization is used to obtain the accurate matching features and spatial mapping relationship of the internal and external damage of the fiber rope. Step 3: Extract damage features and fuse multimodal features from the spatiotemporally registered multimodal detection data obtained in Step 2 to complete intelligent identification and precise positioning of fiber rope damage. Combine damage quantification indicators to achieve fiber rope operation safety assessment and graded early warning.

3. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 2, characterized in that, The specific process of automatically identifying and preprocessing the fiber rope detection data and achieving spatiotemporal synchronous registration in step 2 is as follows: Step 2.1: Perform automatic identification and dynamic tracking of the target fiber rope ROI on the terahertz image and machine vision image, extract the fiber rope detection sub-image in the ROI region, and convert the detection sub-image into a grayscale image; Step 2.2: Perform median filtering and adaptive gamma correction on the grayscale image to obtain the feature-enhanced grayscale image of the fiber rope; Step 2.3: Perform multi-scale SURF feature point extraction on the terahertz image and visual image after feature enhancement within the ROI region, construct a feature descriptor of uniform dimension, and complete the initial matching of fiber rope stability features; Step 2.4: Use the RANSAC algorithm to remove outliers from the initial feature matching pairs to optimize feature matching accuracy; Step 2.5: Perform threshold segmentation and morphological opening and closing operations on the optimized matching image from Step 2.4 to achieve accurate separation of the rope region from the background and smooth the contour of the damaged area. Step 2.6: Based on the image processed in Step 2.5, and combined with the optimized precise feature matching pairs, construct a homography matrix to realize the pixel coordinate space mapping between terahertz internal damage data and visual surface damage data, and complete the precise spatiotemporal synchronous registration of the two types of data in the axial, circumferential and radial dimensions.

4. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 3, characterized in that, Step 2.3, multi-scale SURF feature point extraction, is based on Gaussian filtering and the Hessian matrix. The formula for calculating the Hessian matrix is: ; In the formula, The Hessian matrix; , and denoted as the second-order partial derivatives in the x, xy, and yy directions, respectively.

5. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 3, characterized in that, In step 2.4, when using the RANSAC algorithm to remove outliers, the formula for calculating the required number of iterations is as follows: ; In the formula, k is the required number of iterations; p is the probability that the algorithm successfully samples the "pure interior point set"; e is the proportion of mismatches in the initial feature matching pairs, based on the statistical characteristics of fiber rope bimodal images; and n is the minimum number of feature point pairs required to fit the homography matrix.

6. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 2, characterized in that, Step 3 involves extracting damage features and fusing multimodal features from the spatiotemporally registered multimodal detection data, and completing the damage identification, localization, and safety assessment process. This process includes the following sub-steps: Step 3.1: Perform damage feature submodal extraction on the spatiotemporally registered dual-modal terahertz image and visual image. The terahertz image extracts internal damage features such as internal delamination of the fiber rope, fiber breakage, and matrix voids. The visual image extracts surface damage features such as broken fibers, wear, and corrosion on the fiber rope surface. Step 3.2: Normalize the damage features extracted from the different modes, and use an attention-weighted feature fusion algorithm based on feature importance assessment to assign adaptive weights to the damage features of different modes, and construct a multimodal fusion feature vector of internal and external damage of the fiber rope. Step 3.3: Input the multimodal fusion feature vector into the pre-trained and optimized ResNet-50 bimodal damage classification network to complete the intelligent identification of fiber rope damage type. Combine the location information of ROI dynamic tracking with the spatial mapping relationship established in Step 2 to accurately locate the axial position, circumferential range and radial depth of the damage. Step 3.4: Based on the damage data after identification and positioning, establish a quantitative index system for the severity of damage. Through core parameters such as damage size, defect ratio, number of broken filaments, and wear rate, complete the quantitative classification of the degree of damage to the fiber rope. Step 3.5: Combining the quantitative grading results of fiber rope damage with operating parameters such as the operating load, lifting speed, and running time of the lifting system, construct a fiber rope safety status assessment model. Through this model, first calculate the comprehensive risk value of damage, then derive the safety and reliability coefficient of the fiber rope, comprehensively determine the current operating safety status of the fiber rope, and trigger the corresponding level of early warning signal according to the damage level, and output a fiber rope non-destructive testing and safety assessment report.

7. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 6, characterized in that, In step 3.1, the spatiotemporally registered dual-modal images are fused using Laplacian pyramid multi-scale features. The fusion calculation formula is as follows: And satisfy ; In the formula, The fused Laplacian feature map of the i-th layer and These are the Laplacian pyramid features of the terahertz image and the visual image at the i-th layer, respectively. and These are the fusion weights for the terahertz mode and the visual mode, respectively.

8. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 6, characterized in that, In step 3.3, the loss function of the ResNet-50 bimodal damage classification network is the cross-entropy loss function, calculated as follows: ; In the formula, The loss is the classification loss; N is the batch size. Let i be the true label of the i-th sample; denoted as the predicted probability of the i-th sample; C represents the number of damage categories, including broken fibers, abrasion, delamination, and core breakage. The network minimizes the classification loss through backpropagation, thereby improving the accuracy of fiber rope damage type identification.

9. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 6, characterized in that, Step 3.4 uses the Gray Moment gray-moment damage quantification and grading algorithm to quantify the located damage area. The core calculation formulas include the second-order center gray-moment formula and the normalized gray-moment formula: ; In the formula, m 22 The second-order central gray-level moment of the damaged region characterizes the non-uniformity of gray-level distribution in the damaged region; and The average pixel coordinates of the damaged area. is the pixel grayscale value, and S is the total number of pixels in the damaged area; It is the 0th order grayscale moment (total number of pixels in the damaged area). is the normalized grayscale moment; D is the damage degree quantization value (range 0-1, the larger the value, the more severe the damage), and k is the quantization coefficient.

10. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber rope according to claim 6, characterized in that, The formula for calculating the comprehensive risk value of fiber rope damage in step 3.5, combining the damage type weights, is as follows: ; In the formula, R is the comprehensive risk value of damage, ranging from 0 to 1. The larger the value, the higher the risk of fiber rope operation; ω is the damage degree weighting coefficient, ranging from 0.6 to 0.8, used to characterize the proportion of the impact of the severity of damage on the safety status; D is the quantitative value of damage degree obtained in step 3.4; T is the damage type risk coefficient, which is assigned according to the degree of damage hazard. Severe damage such as rope core breakage and internal delamination is assigned 0.8-1.0, broken wires are assigned 0.3-0.5, and minor damage such as wear is assigned 0.1-0.

2.

11. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber ropes according to claim 1, characterized in that, The terahertz source transmitter operates in the 0.1-1THz frequency band with a spatial resolution better than 0.5mm. The terahertz imager is an area array imaging detector with a maximum sampling frame rate of 60Hz. The multi-camera array monitoring unit includes three high-speed industrial cameras, each with a maximum shooting frame rate of 120Hz and a resolution of 3840×2160. The three high-speed industrial cameras are evenly fixed on a ring-shaped camera bracket using clamps. The camera's field of view completely covers the circumferential surface of the lifting fiber rope, enabling the acquisition of visual images of the fiber rope surface without blind spots. The ring-shaped camera bracket is fixed on the multi-camera array monitoring unit bracket at the sheave, maintaining a preset safe detection distance from the lifting fiber rope to avoid contact with the rope and causing operational interference.

12. The method for damage detection and safety assessment of ultra-deep vertical shaft hoisting fiber rope according to claim 1, characterized in that, The multi-camera array monitoring unit is simplified to two high-speed industrial cameras. It adopts single-scale SURF feature point extraction combined with RANSAC algorithm with fixed number of iterations, simple weighted fusion algorithm and lightweight CNN-like network. It only realizes the axial and circumferential positioning of fiber rope damage and rapid determination of safety status, omitting radial depth fine calibration, micro-defect detection and complex parameter quantization steps.

Images