Steel Wire Rope Monitoring System and Method Based on Wrap-up Flexible Tactile Sensing

Abstract

This invention discloses a wire rope monitoring system and method based on a wraparound flexible tactile sensor, belonging to the field of wire rope condition monitoring technology. It includes a wraparound support, a flexible tactile sensor, a motion positioning component, and a data acquisition and processing module. The latter comprises three units: defect identification and location, intelligent judgment, and severity analysis. The wraparound support has an openable structure to adapt to wire ropes of different specifications. The flexible tactile sensor adheres to its outer surface to collect state data, and the motion positioning component accurately acquires position data. The defect identification and location unit achieves data fusion and initial defect judgment, the intelligent judgment unit completes feature screening and collaborative judgment of defect type and risk level, and the severity analysis unit constructs a health scoring model and pushes maintenance data. This achieves deep fusion of state and position data, improves the accuracy and efficiency of defect identification and location, forms a systematic health assessment system, and realizes the transformation of monitoring data into practical maintenance operations.

Description

Technical Field

[0001] This invention belongs to the field of wire rope condition monitoring technology, specifically relating to a wire rope monitoring system and method based on a wraparound flexible tactile sensor. Background Technology

[0002] Steel wire ropes are core load-bearing components in hoisting, mining, and other fields, and monitoring their operational status is crucial for the safe operation of equipment. While existing monitoring technologies are applied, they suffer from significant shortcomings in data integration, defect identification, and implementation of maintenance, making it difficult to meet actual safety management needs. Therefore, this invention urgently aims to solve the following technical problems: The condition data and spatial location data of the wire rope cannot be effectively integrated, the working condition characteristics are not fully captured, and the accuracy of defect identification and location calibration is insufficient. The feature processing flow for steel wire rope defect identification is poorly designed, resulting in low feature utilization efficiency and unsatisfactory identification efficiency and result reliability. There is no systematic quantitative framework for the health assessment of wire ropes, and the monitoring data cannot be transformed into practical operation and maintenance data, resulting in poor targeting of operation and maintenance work. To address this, we propose a wire rope monitoring system and method based on a wraparound flexible tactile sensor. Summary of the Invention

[0003] The purpose of this invention is to provide a wire rope monitoring system based on a wraparound flexible tactile sensor to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a steel wire rope monitoring system and method based on a wrap-around flexible tactile sensor, comprising: a wrap-around bracket, a flexible tactile sensor, a motion positioning component, and a data acquisition and processing module; The data acquisition and processing module also includes: a defect identification and location unit, a defect intelligent judgment unit, and a defect severity analysis unit; Defect identification and localization unit: calibrates monitoring benchmark points, divides sensing areas and establishes a location benchmark library, and establishes a position mapping relationship between sensing areas and wire ropes; collects wire rope status data, extracts multi-domain joint features and integrates them into feature vectors; determines the defect status of the area through iterative clustering and dynamically updates the cluster center, and builds an integrated dataset of the sensing area. Defect Intelligent Judgment Unit: Constructs a defect area dataset, analyzes the feature dimensions that meet the requirements of comprehensive correlation with defect type and risk level, and obtains a standardized identification feature set after dimensionality reduction; builds an integrated identification model to identify the defect type and risk level of the defect area, forming a three-dimensional integrated data of defect area location-type-risk; Defect Severity Analysis Unit: Build a full-section defect information database, construct a defect type weight assignment database and match the weights of each defect area; calculate the overall health score of the wire rope and match and output the corresponding health level; push the health level and the full-section defect information database to the operation and maintenance personnel to provide data support for targeted governance.

[0005] Preferably, the ring-shaped support includes: a cylindrical outer shell one and a cylindrical outer shell two, one side of the cylindrical outer shell one and the cylindrical outer shell two are hinged together by a hinge, and the end away from the hinge is closed and fixed by a locking member to form an annular cavity that rings around the steel wire rope. The inner arc-shaped surfaces of the first cylindrical shell and the second cylindrical shell are uniformly provided with grooves, and the flexible tactile sensor is fixedly installed inside the grooves. The sensing surface of the flexible tactile sensor is fitted to the outer surface of the steel wire rope. The motion positioning component includes: a positioning bracket, a tension spring, a roller, and an encoder; The positioning bracket is hinged to the upper surface of the cylindrical outer shell, the tension spring is connected between the positioning bracket and the cylindrical outer shell, the roller is rotatably connected to the end of the positioning bracket away from the cylindrical outer shell, the outer circumferential surface of the roller is pressed tightly against the outer surface of the wire rope under the preload of the tension spring, and the encoder is set at the rotating shaft of the roller. The data acquisition and processing module is fixedly mounted on the upper surface of the cylindrical shell. The flexible tactile sensor and the encoder are both electrically connected to the data acquisition and processing module to transmit the acquired sensor data and position data to the data acquisition and processing module.

[0006] Preferably, the specific process of calibrating monitoring benchmark points, dividing sensor areas and establishing a position benchmark database, and establishing the position mapping relationship between sensor areas and wire ropes is as follows: Record the coordinates of the installation reference point of the ring-shaped support relative to the wire rope, and mark it as the initial position for wire rope monitoring; Using the monitoring window of the wraparound support as the core calibration benchmark, the monitoring window is divided into several sensing areas in a circumferential and axial manner and marked sequentially. Measure and record the radial distance and circumferential angle of the geometric center of each sensing area relative to the roller, and establish a sensing area position reference library; obtain the real-time rotation positioning data of the roller, and calculate the real-time axial displacement of the wire rope in combination with the roller circumference; Based on the position reference library, the axial and circumferential coordinates of the steel wire rope corresponding to each sensing area at each monitoring time are calculated. The hardware position of the sensing area is bound to the real-time axial and fixed circumferential position of the steel wire rope, and the position mapping relationship between the two at each monitoring time is established.

[0007] Preferably, the specific process of collecting wire rope state data and extracting multi-domain joint features to integrate them into a feature vector is as follows: Using flexible tactile sensors within the sensing area, raw state data of the outer surface of the wire rope ring at each monitoring time are collected for each sensing area within the monitoring window, including strain, vibration acceleration, surface deformation, and circumferential stress. The data undergoes unified preprocessing. The preprocessed data is then organized to form a standardized raw dataset with dual labels of sensing area location and monitoring time. The time domain, frequency domain, and spatial domain joint features of each sensing area are extracted from the dataset. Finally, the three types of features of the sensing area at each monitoring time are integrated to form the feature vector of the sensing area.

[0008] Preferably, the specific process for building an integrated dataset for the sensor area is as follows: Collect sensor area samples of wire rope defects and normal conditions, calculate the mean of the feature vectors of the two types of samples respectively, and obtain the initial cluster center feature vectors of the defect cluster and the normal cluster. Calculate the Euclidean distance between the current sensing area feature vector and the two cluster centers to determine the defect status of the area; iteratively update the cluster centers until convergence, determine the final judgment result, and use the cluster centers as the initial reference for the next monitoring time. The defect determination results of the area are bound with the identifier, location coordinates, original state data and feature vectors to form an integrated dataset with various labels.

[0009] Preferably, the specific process of constructing a defect area dataset, analyzing the feature dimensions that meet the requirements for comprehensive correlation with defect type and risk level, and obtaining a standardized identification feature set through dimensionality reduction is as follows: Screen out defective sensor areas, extract their integrated datasets and organize them into defective area datasets; collect all types of historical sample data; Following the same process as the defective area, extract joint features from the time domain, frequency domain, and spatial domain to form a full sample feature set with unified dimensions and perform unified preprocessing. The global comprehensive correlation of each feature dimension is analyzed based on the Pearson correlation coefficient analysis method, and features that meet the threshold are selected to form a concise feature set; Principal component analysis is performed on the simplified feature set to calculate the covariance matrix and solve for the eigenvalues ​​and eigenvectors. Principal components with a cumulative variance contribution rate reaching a preset proportion are selected to form a standardized identification feature set.

[0010] Preferably, the specific process of building an integrated identification model to identify the defect type and risk level of defect areas and form three-dimensional integrated data of defect area location-type-risk is as follows: For all samples, a standardized identification feature set is generated, and the corresponding defect type label and risk level label are simultaneously labeled. A multi-output integrated recognition model is constructed. The input of the model is a standardized recognition feature set. The output layer is divided into two paths. The defect type output adopts the Softmax activation function and cross-entropy loss function, while the risk level output adopts the Sigmoid activation function and mean squared error loss function. The total loss function is designed to achieve dual-task collaborative optimization. The model is iteratively trained with a sample dataset labeled with two labels until the model's judgment accuracy on the validation set reaches the preset standard. Obtain the standardized identification feature set of each defect area from the defect area dataset at the current monitoring time, input it into the trained integrated identification model, and the model synchronously outputs the defect type and risk level of the corresponding defect area. By binding the defect type, risk level, and the location coordinates and original feature data of the corresponding sensing area, a three-dimensional integrated data of defect area location, type, and risk is formed.

[0011] Preferably, a full-section defect information database is established, a defect type weight assignment database is constructed, and the weights of each defect area are matched; the specific process for calculating the overall health score of the wire rope is as follows: Construct a defect type weight assignment library, and set corresponding weights according to the degree of influence of each type of defect on the load-bearing capacity of the wire rope. The greater the degree of influence, the higher the weight value. Based on the full-section defect information database and the defect type weight assignment database, corresponding weight assignments are matched for each defect area; an overall health scoring model for the wire rope is constructed, which incorporates the total number of defects, the defect type weight assignments for each defect area, the preset basic operating parameter correction coefficients, as well as the maximum value of the defect type weight assignments and the maximum value of the risk level into the model calculation to obtain the overall health score of the wire rope.

[0012] Preferably, the corresponding health level is matched and output; the specific process of pushing the health level and the full-segment defect information database to the operation and maintenance personnel terminal to provide data support for targeted governance is as follows: The steel wire rope is divided into several health levels, and each health level corresponds to a health score range. The current health score of the wire rope is matched with the health score range corresponding to all wire rope health levels, and the corresponding health level of the wire rope is output. The health level and the full-section defect information database are pushed to the operation and maintenance personnel's terminal. The operation and maintenance personnel can determine the operation and maintenance priority based on the health level, and implement targeted management of the wire rope by combining the data such as defect location, type and risk level in the full-section defect information database.

[0013] Compared with the prior art, the beneficial effects of the present invention are: (1) The wire rope monitoring system and method based on the ring-shaped flexible tactile sensing adapts to wire ropes of different specifications through the opening and closing structure of the ring-shaped bracket. The flexible tactile sensor in the inner groove of the bracket is attached to the outer surface of the wire rope to collect state data. The motion positioning component composed of the tension spring pre-tightened roller and the encoder obtains accurate position data. Then, the defect identification and positioning unit builds a position benchmark library, extracts the joint features of time domain, frequency domain and spatial domain and iteratively clusters to determine the defect. This realizes the deep integration of state data and spatial position data, comprehensively captures the working condition characteristics of the wire rope, and greatly improves the accuracy of defect identification and position calibration.

[0014] (2) The wire rope monitoring system and method based on the wrap-around flexible tactile sensing constructs a defect area dataset, selects feature dimensions that are highly correlated with defect type and risk level based on Pearson correlation coefficient analysis, obtains a standardized identification feature set through principal component analysis for dimensionality reduction, and builds a multi-output integrated identification model to achieve collaborative judgment of defect type and risk level, optimizes the feature processing process, improves feature utilization efficiency, and effectively improves the efficiency and reliability of defect judgment.

[0015] (3) The wire rope monitoring system and method based on the wrap-around flexible tactile sensing builds a full-section defect information database, constructs a weighted value database based on the degree of influence of defects on the wire rope bearing capacity, constructs an overall health scoring model of the wire rope in combination with the equipment operating conditions and matches and outputs the corresponding health level, and pushes the health level and the full-section defect information database to the operation and maintenance personnel, forming a systematic wire rope health quantitative assessment system, realizing the effective transformation of monitoring data into operation and maintenance practice basis, and significantly improving the pertinence and effectiveness of wire rope operation and maintenance work. Attached Figure Description

[0016] Figure 1 This is a flowchart of the present invention; Figure 2 This is a three-dimensional structural diagram of the wrap-around support of the present invention; Figure 3 This is a schematic diagram of the assembly structure of the wrap-around bracket of the present invention.

[0017] In the diagram: 11. Cylindrical outer shell one; 12. Cylindrical outer shell two; 101. Groove; 13. Data acquisition and processing module; 14. Positioning bracket; 15. Tension spring; 16. Roller. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only 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.

[0019] Example 1; Please see Figure 1-3 The present invention provides a wire rope monitoring system based on a wrap-around flexible tactile sensor, comprising: a wrap-around bracket, a flexible tactile sensor, a motion positioning component, and a data acquisition and processing module; The ring-shaped support includes: a cylindrical outer shell 11 and a cylindrical outer shell 2 12. One side of the cylindrical outer shell 11 and the cylindrical outer shell 2 12 are hinged together by a hinge, and the end away from the hinge is closed and fixed by a locking member to form an annular cavity that hugs the steel wire rope. The inner arc-shaped surfaces of cylindrical shell 11 and cylindrical shell 2 are uniformly provided with grooves 101. A flexible tactile sensor is fixedly installed inside the grooves 101, and the sensing surface of the flexible tactile sensor is fitted to the outer surface of the steel wire rope. The motion positioning assembly includes: a positioning bracket 14, a tension spring 15, a roller 16, and an encoder; the positioning bracket 14 is hinged to the upper surface of the cylindrical outer shell 11, the tension spring 15 is connected between the positioning bracket 14 and the cylindrical outer shell 11, the roller 16 is rotatably connected to the end of the positioning bracket 14 away from the cylindrical outer shell 11, the outer circumferential surface of the roller 16 is pressed tightly against the outer surface of the wire rope under the preload of the tension spring 15, and the encoder is set at the pivot of the roller 16; The data acquisition and processing module 13 is fixedly mounted on the upper surface of the cylindrical shell 11. The flexible tactile sensor and the encoder are both electrically connected to the data acquisition and processing module 13 to transmit the acquired sensor data and position data to the data acquisition and processing module 13.

[0020] It should be noted that the real-time and accurate monitoring of the wire rope is achieved through the coordinated operation of various components. The overall working process is as follows: First, the cylindrical outer shell 11 and the cylindrical outer shell 12 are opened through the hinge and fitted onto the outside of the wire rope to be monitored. Then, the two are closed and fixed by the locking device, so that the sensing surface of the flexible tactile sensor in the groove 101 is in close contact with the outer surface of the wire rope. At the same time, the preload of the tension spring 15 drives the positioning bracket 14 to press the outer circumference of the roller 16 into close contact with the outer surface of the wire rope, completing the assembly and bonding preparation before monitoring; when the wire rope moves axially... The friction between the wire rope and the roller 16 drives the roller 16 to rotate synchronously. The encoder at the shaft of the roller 16 records the rotation data in real time and transmits it to the data acquisition and processing module 13. At the same time, the flexible tactile sensor collects the state data of the wire rope, such as strain, vibration, and surface deformation, in real time and also transmits it to the data acquisition and processing module 13. The data acquisition and processing module 13 integrates, processes, and performs preliminary analysis on the received sensor data and position data to bind the state of the wire rope with the corresponding axial position, providing data support for the identification and location of wire rope defects. The ring-shaped bracket adopts an openable structure design with hinges and locking parts. The disassembly and assembly of the cylindrical outer shell 11 and the cylindrical outer shell 2 12 are convenient, realizing the rapid disassembly and assembly and stable fixation of the monitoring system on the wire rope, which greatly improves the convenience of installation and maintenance of the device and is suitable for monitoring needs of wire ropes of different specifications. The flexible tactile sensor is directly attached to the outer surface of the wire rope, which can accurately capture the minute changes in the state of the wire rope. Combined with the design of grooves 101 evenly distributed on the inner arc surface of cylindrical shell 11 and cylindrical shell 22, it can realize comprehensive monitoring of the circumference of the wire rope, and improve the sensitivity and comprehensiveness of the wire rope condition monitoring. The motion positioning component ensures that the roller 16 and the wire rope are continuously pressed and adhered by the preload of the tension spring 15, effectively avoiding slippage.

[0021] The data acquisition and processing module also includes: a defect identification and location unit, a defect intelligent judgment unit, and a defect severity analysis unit; The defect identification and localization unit: This unit calibrates monitoring benchmark points, divides sensing areas, and establishes a location benchmark database; it also establishes a position mapping relationship between sensing areas and wire ropes; it collects wire rope status data, extracts multi-domain joint features, and integrates them into a feature vector; it determines the defect status of the area through iterative clustering and dynamically updates the cluster centers, thus building an integrated dataset for the sensing areas. The specific process is as follows: Record the coordinates of the installation reference point of the ring-shaped support relative to the wire rope. This benchmark point is marked as the initial location for wire rope monitoring; Using the monitoring window of the wraparound support as the core calibration reference, the monitoring window is divided into several sensing areas in a circumferentially evenly divided and axially segmented manner, and marked as follows. i is the label of the sensor area, and M is the total number of sensor areas; For each sensor area Measure and record the radial distance of its geometric center relative to the rotation axis of the roller (16). Circumferential angle To form a location reference library for the sensing area ; Acquire real-time rotational positioning data of the roller (16), including rotational angular velocity. Total number of rotations Combining the circumference L of the roller (16), using the formula: The real-time axial displacement of the wire rope is obtained. ;in, This represents the cumulative angular displacement of the roller (16) from the initial time 0 to time t; This indicates the axial displacement of the wire rope corresponding to each radian (unit angle) rotated by the roller; Based on the location reference library of sensor areas For each sensor area in the library Using the formula: This yields the axial coordinates of the wire rope for each sensing area at each monitoring time. Circumferential coordinates ; The hardware location of the sensing area is bound to the real-time axial position and fixed circumferential position of the wire rope to establish a mapping relationship between the sensing area and the actual position of the wire rope at each monitoring time. For each sensing area within the monitoring window, the raw state data of the outer surface of the wire rope ring at each monitoring moment is acquired through a flexible tactile sensor within the area, including strain. Vibration acceleration Surface deformation Circumferential stress And perform unified preprocessing on the collected raw state data; The raw state data of each sensor area at each monitoring time after preprocessing are organized to form a standardized raw dataset with sensor area location labels and monitoring time labels. In the standardized raw dataset, each sensor area-monitoring time combination corresponds to the actual axial position information, fixed circumferential position information and preprocessed raw state data of the wire rope. For the raw state data of each sensing patch in the standardized raw dataset, joint features in the time domain, frequency domain, and spatial domain are extracted, specifically as follows: Time-domain characteristics include: mean strain, root mean square vibration acceleration, and peak surface deformation. mean strain The calculation formula is: Where T is the duration of a single monitoring session; Root mean square of vibration acceleration The calculation formula is: ; Peak surface deformation The calculation formula is: ; Frequency domain characteristics include: center frequency and main band energy; Center frequency The calculation process is as follows: for vibration acceleration Perform a Fast Fourier Transform on the frequency domain sequence in complex form, and extract the absolute value of the modulus of the complex sequence to obtain the frequency domain amplitude. Using the formula: , The sampling frequency; Main frequency band energy The calculation formula is: , Main frequency bandwidth; Spatial characteristics include: the circumferential stress difference between adjacent sensing areas and the sensor response consistency coefficient within the sensing area; Circumferential stress difference between adjacent sensing areas The calculation formula is: ; Sensor response consistency coefficient within the sensing area The calculation formula is: Where k is the sensor number within the sensing area, and K is the total number of sensors within the sensing area. The response value of the k-th sensor in the sensing area is obtained by extracting the joint features of time domain, frequency domain, and spatial domain (mean strain, root mean square of vibration acceleration, peak value of surface deformation, etc.) of each sensor, normalizing each feature, and then calculating its Euclidean norm as the response value of the sensor. This is the average sensor response within the sensing area (calculated from the response values ​​of each sensor within the area). The time-domain features, frequency-domain features, and spatial-domain features of each sensing area at the monitoring time are integrated to form the feature vector of the sensing area. For different types of defects and normal states of steel wire rope, several corresponding sensor area samples were collected. Each sample contains a complete feature vector (covering features of each dimension corresponding to the time domain, frequency domain, and spatial domain), and the number of samples meets the statistical validity requirements for model training. All sensor area samples corresponding to wire rope defects are grouped into one category and labeled as defect clusters. The mean of each dimension of the feature vector of all samples in the defect cluster is calculated to obtain the initial cluster center feature vector of the defect cluster. All sensor area samples corresponding to the normal state of the wire rope are grouped into one category and labeled as the normal cluster. The mean of each dimension of the feature vector of all samples in the normal cluster is calculated to obtain the initial cluster center feature vector of the normal cluster. The initial cluster center feature vectors of defect clusters and normal clusters are used as the initial reference feature vectors for determining the category of each sensing area at the current monitoring time. For each sensing area at the current monitoring time, calculate the Euclidean distance between its feature vector and the feature vector of the initial cluster center of the defect cluster and the normal cluster (the parameters have been normalized and dimensionless) to obtain the Euclidean distance of the sensing area relative to the defect cluster and the Euclidean distance relative to the normal cluster. If the Euclidean distance between the sensing area and the defect cluster is less than the Euclidean distance between the sensing area and the normal cluster, the sensing area is determined to have a defect; otherwise, the sensing area is determined to be in normal condition. Based on the judgment results of all sensor areas at the current monitoring time, the sensor area sets corresponding to the defect cluster and the normal cluster are reorganized, and the mean of each dimension of the feature vector of all sensor areas in the two clusters is calculated respectively. These are used as the feature vector of the defect cluster cluster center and the feature vector of the normal cluster cluster center for the new round of judgment. Calculate the absolute value of the difference between the feature vector of the new cluster center of the defective cluster and the feature vector of the corresponding cluster center of the previous round in each dimension. If the absolute value of the difference in all dimensions of the defective cluster and the normal cluster is less than the preset convergence threshold (i.e. the positions of the two cluster centers are basically unchanged), then stop the iteration. If the convergence condition is not met, repeat the steps of calculating the Euclidean distance → determining the category → updating the cluster center feature vector until the convergence condition is met or the preset number of iterations is reached. The final determination results of each sensor area after the final iteration are used as the status determination results of each sensor area at the current monitoring time. The results include: whether there are defects in the sensor area and the cluster category to which it belongs. Meanwhile, the final defect cluster center feature vector and normal cluster center feature vector are used as the initial cluster center feature vector for the status determination of each sensing area at the next monitoring time, so as to realize the dynamic iterative update of the cluster center. The defect determination results of the sensing area are bound with the corresponding sensing area identifier, location coordinates, original state data, and feature vector to form an integrated dataset with sensing area identifier, location information, original state, feature vector, and defect determination label.

[0022] It should be noted that by establishing a reference library for the location of the sensor area, combining the position mapping relationship with the roller rotation positioning data, and coordinating the fit between the roller and the wire rope, it is helpful to achieve reliable calibration of the defect location, improve the problem of the disconnect between the status data and the actual location in traditional monitoring, and provide spatial support for subsequent defect location and maintenance. By adopting a circumferentially evenly divided and axially segmented sensing area division method, and combining it with the flexible tactile sensor to collect data on the annular outer surface of the wire rope, a relatively comprehensive state data acquisition can be achieved. At the same time, by eliminating interference noise through a unified preprocessing process, it is beneficial to ensure the standardization and integrity of the original data, laying a data foundation for subsequent feature extraction and defect judgment. A multi-dimensional joint feature extraction system encompassing the time, frequency, and spatial domains is constructed to overcome the limitations of single-dimensional feature representation. By integrating multi-dimensional features to form feature vectors, the operating state of wire ropes can be quantitatively characterized at multiple levels, which helps to improve the feature's ability to identify and distinguish different types of defects, and provides feature-level support for defect judgment. The dynamic iterative clustering analysis method avoids the inherent limitations of the traditional fixed threshold method. By iteratively optimizing the cluster centers in real time, the model can adapt to different operating conditions and defect change patterns of wire ropes. It can not only identify small defects in the early stage, but also reduce the impact of changes in operating conditions on the judgment results, which is conducive to improving the accuracy and adaptability of defect judgment. Establishing an integrated dataset binding mechanism binds sensor area identifiers, location information, raw state data, feature vectors, and defect judgment results, enabling unified data management and direct reuse. This reduces secondary data acquisition and processing operations in subsequent units, improves the overall efficiency of the data acquisition and processing module, and reduces system computational overhead. Using the cluster center of the current monitoring cycle as the initial reference for the next cycle enables iterative inheritance of the cluster center, eliminating the need for repeated model training and data reprocessing. This forms a lightweight real-time monitoring optimization method, which helps reduce the system's computational load and ensures the continuity and efficiency of wire rope condition monitoring.

[0023] The intelligent defect determination unit constructs a defect area dataset, analyzes the feature dimensions that meet the requirements for comprehensive correlation with defect type and risk level, and obtains a standardized identification feature set through dimensionality reduction; it then builds an integrated identification model to identify the defect type and risk level of the defect area, forming a three-dimensional integrated data of defect area location, type, and risk. The specific process is as follows: For each monitoring moment, the sensor areas marked as having defects are selected, and the integrated dataset (including location labels, raw state data, and feature vectors) corresponding to all defective sensor areas is extracted and organized to form the defective sensor area dataset for the current monitoring moment. Collect historical sample data covering all defect types, risk levels, and normal states. Following a process completely consistent with the defect area, extract joint features in the time domain, frequency domain, and spatial domain to form a full sample feature set with unified dimensions. Time-domain characteristics: mean strain, root mean square vibration acceleration, peak surface deformation; Frequency domain characteristics: center frequency, main frequency band energy; Spatial characteristics: circumferential stress difference between adjacent sensing areas, and sensor response consistency coefficient within the sensing area; The sample feature set is preprocessed uniformly, including: removing outliers (screening according to preset outlier judgment criteria and using linear interpolation to fill in missing values), and standard normalization and dimensionless processing. For each feature dimension in the full sample feature set, the following calculations are performed based on Pearson correlation coefficient analysis: The degree of comprehensive correlation between feature dimensions and each defect type, A1 (absolute value form, reflecting the ability of features to distinguish defect types). The comprehensive correlation between feature dimensions and the risk level of each defect type, A2 (absolute value form, reflecting the ability of features to quantify risk level). Using the formula: AZ=A1×b1+A2×b2, we can obtain the global comprehensive correlation degree AZ (the comprehensive contribution of features to the dual-label determination), where b1 and b2 are preset weight coefficients; A global comprehensive correlation threshold is preset, and feature dimensions with a global comprehensive correlation greater than or equal to the corresponding preset threshold are selected and organized to form a simplified feature set. Principal component analysis was applied to the simplified feature set, including: Calculate the covariance matrix of the simplified feature set (to uncover the intrinsic linear correlation between features). Solve for the eigenvalues ​​and corresponding eigenvectors of the covariance matrix, sort them by eigenvalues ​​from largest to smallest, and quantify the information carrying capacity of each principal component; Select the top R principal components whose cumulative variance contribution rate reaches a preset proportion to form a standardized identification feature set; Obtain the standardized identification feature set of all samples, and simultaneously label each sample with the corresponding defect type label and risk level label; A multi-output integrated identification model is constructed. The model input is a standardized identification feature set, and the output layer is divided into two paths, corresponding to defect type and risk level, respectively. Defect type output: Softmax activation function is used, and cross-entropy loss is employed. , in, Cross-entropy loss for defect type classification; This represents the total number of training samples; The true defect type label for the nth sample (using one-hot encoding, with a value of 1 for the true category position and 0 for the other category positions); This is the probability distribution of the defect type predicted by the model for the nth sample (output after the Softmax activation function, the sum of the probabilities of all categories is 1); Risk level output: Sigmoid activation function is used, and the loss function is mean squared error loss. , in, The mean squared error loss for risk level regression; This represents the true risk level label for the nth sample. This represents the risk level rating predicted by the model for the nth sample. The squared difference between the true risk level and the predicted risk level of the nth sample is used to measure the prediction error of the regression task. The total loss function is: , in, The optimization objective for model training is to achieve collaborative optimization of the two tasks by minimizing the total loss; Preset weighting coefficients; Train the model by inputting a sample dataset with double labels, and iterate the training until the model’s judgment accuracy on the validation set reaches the preset standard, thus obtaining an integrated recognition model. Based on the defect area dataset at the current monitoring time, a standardized identification feature set for each defect area is obtained, and the standardized identification feature set of the defect area is input into the trained integrated identification model. The model simultaneously outputs the defect type and risk level of the corresponding defect area. By binding the defect type and risk level of the defect area with the location coordinates and original feature data of the sensing area, a three-dimensional integrated data of location, type and risk is formed.

[0024] It should be noted that preprocessing operations such as removing outliers, using linear interpolation to fill in missing values, and standard normalization to remove dimensions on the sample feature set can effectively reduce the impact of outlier data interference and dimensional differences on feature effectiveness, improve the overall quality of the sample feature set, and lay a good data foundation for subsequent feature selection and model training. The global comprehensive correlation of each feature dimension is calculated based on the Pearson correlation coefficient analysis method. High correlation features are selected by preset threshold to form a simplified feature set. Redundant features with low contribution to the determination of defect type and risk level can be eliminated, reducing the computational burden in the subsequent model training and inference process, while strengthening the supporting role of core features in the determination of defect dual labels. Principal component analysis is used to reduce the dimensionality of the simplified feature set. By selecting principal components with a cumulative variance contribution rate of a preset ratio to form standardized recognition features, the feature dimensionality can be further reduced while retaining the main information of the features, simplifying the model structure and effectively improving the training efficiency and actual inference efficiency of the model. An integrated identification model with multiple outputs is constructed, which integrates two tasks: defect type classification and risk level regression. By designing a total loss function, collaborative optimization training of the two tasks is achieved. Compared with the method of independently constructing two task models, information sharing between the two tasks can be realized, reducing redundant operations in model construction and training, while improving the consistency of defect type and risk level judgment results. The defect type and risk level are bound to the sensor area location coordinates and original feature data to form a three-dimensional integrated data of location, type and risk. This enables the systematic integration and management of defect-related information, providing comprehensive and complete data support for the subsequent analysis of the overall health of the wire rope. At the same time, it enables seamless connection of data between units and improves the data flow efficiency of the entire data acquisition and processing module.

[0025] Defect Severity Analysis Unit: This unit establishes a comprehensive defect information database, constructs a defect type weight assignment library, and matches the weights of each defect area; calculates the overall health score of the wire rope and outputs the corresponding health level; and pushes the health level and the comprehensive defect information database to the maintenance personnel's end to provide data support for targeted governance. The specific process is as follows: When the entire section of the wire rope is monitored, the three-dimensional integrated data of all defect areas of the entire section of the wire rope is collected to form a complete defect information database, which includes all information such as defect location, type, risk level, and standardized identification features, providing complete data support for overall health assessment and segmented analysis. Construct a defect type weight assignment library, in which each defect type has a corresponding weight assignment (set according to the degree of influence of different defect types on the load-bearing capacity of wire rope, the greater the influence, the higher the weight value). Based on the full-section defect information database and defect type weight assignment database of the wire rope, the weight assignment of each defect area is matched. Constructing an overall health scoring model for steel wire ropes: , in, Assess the health of the wire rope; The total number of defects is r, where r is the label of the defect area. Assign a weight to the defect type of the r-th defect region; D is the preset basic operating parameter correction coefficient (which can be calculated from the design service life of the wire rope, the design cumulative running time, and the actual operating parameters; the greater the actual operating loss, the smaller the D value). The maximum value assigned to the weights of all defect types; This represents the maximum value across all risk levels. The steel wire rope is divided into several health levels, and each health level corresponds to a health score range. The larger the upper and lower limits of the health score range, the better the health of the steel wire rope. The current health score of the wire rope is matched with the health score range corresponding to the health levels of all wire ropes, and the corresponding health level of the wire rope is output. Then, the health level and the full-section defect information database are pushed to the operation and maintenance personnel. Maintenance personnel can determine maintenance priorities based on health levels and implement targeted management of wire ropes by combining data such as defect location, type, and risk level in the full-section defect information database.

[0026] It should be noted that the three-dimensional integrated data of the entire defect area is compiled to form a full-section defect information database, which integrates information on defect location, type, risk level and other dimensions. This provides systematic and complete data support for the overall health assessment and segmented analysis of the wire rope, and ensures the comprehensiveness of subsequent assessment work. A weighted value library based on the degree of impact of defects on the load-bearing capacity of wire ropes was constructed, realizing differentiated quantification of different defect types, so that the health score can fit the actual safety influencing factors of wire ropes, and improving the rationality and practical reference of the health score. The system presets health levels and corresponding scoring ranges, transforming quantitative health scores into intuitive level determinations. This makes it easier for maintenance personnel to quickly identify the health status of wire ropes, reduces the difficulty of data interpretation, and improves the practicality and operability of the assessment results. The health level and the full-section defect information database are pushed to the operation and maintenance personnel's terminal simultaneously. The operation and maintenance personnel can determine the operation and maintenance priority based on the health level and implement targeted management in combination with the full-dimensional defect information. This realizes the effective transformation of monitoring data into operation and maintenance actions, making operation and maintenance work more targeted, helping to reduce the waste of resources caused by blind operation and maintenance, and improving the overall efficiency of wire rope operation and maintenance work.

[0027] A wire rope monitoring method based on wraparound flexible tactile sensing includes: Step 1: Calibrate monitoring benchmark points, divide sensing areas and establish a location benchmark library, and build a position mapping relationship between sensing areas and wire ropes; collect wire rope status data, extract multi-domain joint features and integrate them into feature vectors; determine the defect status of the area through iterative clustering and dynamically update the cluster centers to build an integrated dataset of the sensing areas. Step 2: Construct a defect area dataset, analyze the feature dimensions that meet the requirements for comprehensive correlation with defect type and risk level, and obtain a standardized identification feature set after dimensionality reduction; build an integrated identification model to identify the defect type and risk level of the defect area, forming a three-dimensional integrated data of defect area location-type-risk. Step 3: Build a full-section defect information database, construct a defect type weight assignment database and match the weights of each defect area; calculate the overall health score of the wire rope and match and output the corresponding health level; push the health level and the full-section defect information database to the operation and maintenance personnel to provide data support for targeted governance.

[0028] Although embodiments of the invention have been shown and described, 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 wire rope monitoring system based on a wraparound flexible tactile sensor, characterized in that, include: Wrap-around bracket, flexible tactile sensor, motion positioning component and data acquisition and processing module; The data acquisition and processing module also includes: a defect identification and location unit, a defect intelligent judgment unit, and a defect severity analysis unit; Defect identification and localization unit: calibrates monitoring benchmark points, divides sensing areas and establishes a location benchmark library, and establishes a position mapping relationship between sensing areas and wire ropes; collects wire rope status data, extracts multi-domain joint features and integrates them into feature vectors; determines the defect status of the area through iterative clustering and dynamically updates the cluster center, and builds an integrated dataset of the sensing area. Defect Intelligent Judgment Unit: Constructs a defect area dataset, analyzes the feature dimensions that meet the requirements of comprehensive correlation with defect type and risk level, and obtains a standardized identification feature set after dimensionality reduction; builds an integrated identification model to identify the defect type and risk level of the defect area, forming a three-dimensional integrated data of defect area location-type-risk; Defect Severity Analysis Unit: Build a full-section defect information database, construct a defect type weight assignment database and match the weights of each defect area; calculate the overall health score of the wire rope and match and output the corresponding health level; push the health level and the full-section defect information database to the operation and maintenance personnel to provide data support for targeted governance.

2. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 1, characterized in that: The ring-shaped support includes: a cylindrical outer shell one (11) and a cylindrical outer shell two (12). One side of the cylindrical outer shell one (11) and the cylindrical outer shell two (12) are hinged together by a hinge, and the end away from the hinge is closed and fixed by a locking member to form an annular cavity that hugs the steel wire rope. The inner arc-shaped surfaces of the first cylindrical shell (11) and the second cylindrical shell (12) are uniformly provided with grooves (101), and the flexible tactile sensor is fixedly installed inside the grooves (101). The sensing surface of the flexible tactile sensor is attached to the outer surface of the steel wire rope. The motion positioning component includes: a positioning bracket (14), a tension spring (15), a roller (16), and an encoder; The positioning bracket (14) is hinged to the upper surface of the cylindrical outer shell (11), the tension spring (15) is connected between the positioning bracket (14) and the cylindrical outer shell (11), the roller (16) is rotatably connected to the end of the positioning bracket (14) away from the cylindrical outer shell (11), the outer circumferential surface of the roller (16) is pressed against the outer surface of the wire rope under the preload of the tension spring (15), and the encoder is set at the rotating shaft of the roller (16); The data acquisition and processing module (13) is fixedly installed on the upper surface of the cylindrical shell (11). The flexible tactile sensor and the encoder are both electrically connected to the data acquisition and processing module (13) to transmit the acquired sensor data and position data to the data acquisition and processing module (13).

3. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 2, characterized in that: The specific process of calibrating monitoring benchmark points, dividing sensor areas and establishing a location benchmark database, and establishing the position mapping relationship between sensor areas and wire ropes is as follows: Record the coordinates of the installation reference point of the ring-shaped support relative to the wire rope, and mark it as the initial position for wire rope monitoring; Using the monitoring window of the wraparound support as the core calibration benchmark, the monitoring window is divided into several sensing areas in a circumferential and axial manner and marked sequentially. Measure and record the radial distance and circumferential angle of the geometric center of each sensing area relative to the roller (16) to establish a sensing area position reference library; obtain the real-time rotation positioning data of the roller (16) and calculate the real-time axial displacement of the wire rope in combination with the circumference of the roller (16); Based on the position reference library, the axial and circumferential coordinates of the steel wire rope corresponding to each sensing area at each monitoring time are calculated. The hardware position of the sensing area is bound to the real-time axial and fixed circumferential position of the steel wire rope, and the position mapping relationship between the two at each monitoring time is established.

4. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 3, characterized in that: The specific process of collecting wire rope status data and extracting multi-domain joint features to integrate them into a feature vector is as follows: Using flexible tactile sensors within the sensing area, raw state data of the outer surface of the wire rope ring at each monitoring time are collected for each sensing area within the monitoring window, including strain, vibration acceleration, surface deformation, and circumferential stress. The data undergoes unified preprocessing. The preprocessed data is then organized to form a standardized raw dataset with dual labels of sensing area location and monitoring time. The time domain, frequency domain, and spatial domain joint features of each sensing area are extracted from the dataset. Finally, the three types of features of the sensing area at each monitoring time are integrated to form the feature vector of the sensing area.

5. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 4, characterized in that: The specific process of building an integrated dataset for the sensor area is as follows: Collect sensor area samples of wire rope defects and normal conditions, calculate the mean of the feature vectors of the two types of samples respectively, and obtain the initial cluster center feature vectors of the defect cluster and the normal cluster. Calculate the Euclidean distance between the current sensing area feature vector and the two cluster centers to determine the defect status of the area; iteratively update the cluster centers until convergence, determine the final judgment result, and use the cluster centers as the initial reference for the next monitoring time. The defect determination results of the area are bound with the identifier, location coordinates, original state data and feature vectors to form an integrated dataset with various labels.

6. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 5, characterized in that: The specific process of constructing a defect area dataset, analyzing the feature dimensions that meet the requirements for comprehensive correlation with defect type and risk level, and obtaining a standardized identification feature set through dimensionality reduction is as follows: Screen out defective sensor areas, extract their integrated datasets and organize them into defective area datasets; collect all types of historical sample data; Following the same process as the defective area, extract joint features from the time domain, frequency domain, and spatial domain to form a full sample feature set with unified dimensions and perform unified preprocessing. The global comprehensive correlation of each feature dimension is analyzed based on the Pearson correlation coefficient analysis method, and features that meet the threshold are selected to form a concise feature set; Principal component analysis is performed on the simplified feature set to calculate the covariance matrix and solve for the eigenvalues ​​and eigenvectors. Principal components with a cumulative variance contribution rate reaching a preset proportion are selected to form a standardized identification feature set.

7. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 6, characterized in that: The specific process of building an integrated identification model to identify the defect type and risk level of defect areas and form three-dimensional integrated data of defect area location, type, and risk is as follows: For all samples, a standardized identification feature set is generated, and the corresponding defect type label and risk level label are simultaneously labeled. A multi-output integrated recognition model is constructed. The input of the model is a standardized recognition feature set. The output layer is divided into two paths. The defect type output adopts the Softmax activation function and cross-entropy loss function, while the risk level output adopts the Sigmoid activation function and mean squared error loss function. The total loss function is designed to achieve dual-task collaborative optimization. The model is iteratively trained with a sample dataset labeled with two labels until the model's judgment accuracy on the validation set reaches the preset standard. Obtain the standardized identification feature set of each defect area from the defect area dataset at the current monitoring time, input it into the trained integrated identification model, and the model synchronously outputs the defect type and risk level of the corresponding defect area. By binding the defect type, risk level, and the location coordinates and original feature data of the corresponding sensing area, a three-dimensional integrated data of defect area location, type, and risk is formed.

8. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 7, characterized in that: The process of establishing a comprehensive defect information database, constructing a defect type weight assignment database, and matching the weights of each defect area is as follows: The specific process for calculating the overall health score of the wire rope is as follows: Construct a defect type weight assignment library, and set corresponding weights according to the degree of influence of each type of defect on the load-bearing capacity of the wire rope. The greater the degree of influence, the higher the weight value. Based on the full-section defect information database and the defect type weight assignment database, corresponding weight assignments are matched for each defect area; an overall health scoring model for the wire rope is constructed, which incorporates the total number of defects, the defect type weight assignments for each defect area, the preset basic operating parameter correction coefficients, as well as the maximum value of the defect type weight assignments and the maximum value of the risk level into the model calculation to obtain the overall health score of the wire rope.

9. The wire rope monitoring system based on a wraparound flexible tactile sensor according to claim 8, characterized in that: The specific process of matching the output to the corresponding health level and pushing the health level and the full-segment defect information database to the operations and maintenance personnel to provide data support for targeted governance is as follows: The steel wire rope is divided into several health levels, and each health level corresponds to a health score range. The current health score of the wire rope is matched with the health score range corresponding to all wire rope health levels, and the corresponding health level of the wire rope is output. The health level and the full-section defect information database are pushed to the operation and maintenance personnel's terminal. The operation and maintenance personnel can determine the operation and maintenance priority based on the health level, and implement targeted management of the wire rope by combining the data such as defect location, type and risk level in the full-section defect information database.

10. A wire rope monitoring method based on a wraparound flexible tactile sensor, applied to the wire rope monitoring system based on a wraparound flexible tactile sensor as described in any one of claims 1-9, characterized in that, include: Step 1: Calibrate monitoring benchmark points, divide sensor areas and establish a position benchmark library, and establish a position mapping relationship between sensor areas and steel wire ropes; Collect wire rope status data, extract multi-domain joint features and integrate them into feature vectors; determine the defect status of the area through iterative clustering and dynamically update the cluster centers to build an integrated dataset of the sensing area; Step 2: Construct a defect area dataset, analyze the feature dimensions that meet the requirements for comprehensive correlation with defect type and risk level, and obtain a standardized identification feature set after dimensionality reduction; build an integrated identification model to identify the defect type and risk level of the defect area, forming a three-dimensional integrated data of defect area location-type-risk. Step 3: Build a full-section defect information database, construct a defect type weight assignment library, and match the weights of each defect area; Calculate the overall health score of the wire rope and match the corresponding health level; push the health level and the full-section defect information database to the operation and maintenance personnel to provide data support for targeted governance.

Images