Intelligent detection method and device for deformation of cage steel wire rope based on image recognition

Abstract

The application discloses a cage steel wire rope deformation intelligent detection method and device based on image recognition and relates to the technical field of image recognition.The method comprises the following steps: based on the starting and stopping state of the cage and the recent working condition parameters, acquiring a collection frequency, collecting and acquiring a rope diameter image sequence and a lay distance image sequence; respectively performing defect identification, acquiring rope diameter defect parameters and lay distance defect parameters; comprehensively processing the rope diameter image sequence and the lay distance image sequence, performing surface anomaly identification, and acquiring surface anomaly parameters; based on the rope diameter defect parameters, the lay distance defect parameters and the surface anomaly parameters, combining the starting and stopping state of the cage and the recent working condition parameters, judging the cage steel wire rope deformation condition, acquiring a cage steel wire rope deformation identification result, and adjusting the collection frequency to perform the cage steel wire rope deformation intelligent detection in the next identification cycle.The application solves the technical problems of poor reliability and efficiency of the cage steel wire rope deformation condition detection in the prior art.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and specifically to an intelligent detection method and device for deformation of steel wire rope in cages based on image recognition. Background Technology

[0002] As a critical load-bearing component of mine hoisting systems, the health of cage wire ropes directly impacts mine safety and personnel safety. Under prolonged exposure to alternating loads and friction, wire ropes are prone to various deformations and damages, including localized wear in the rope diameter and abnormal lay pitch. Existing cage wire rope testing technologies often suffer from fixed frequencies and insufficient reliability verification, resulting in poor robustness and accuracy in complex field environments. Summary of the Invention

[0003] This application provides an intelligent detection method and device for deformation of wire rope in cages based on image recognition, which is used to address the technical problems of poor reliability and efficiency in detecting deformation of wire rope in cages in the prior art.

[0004] In view of the above problems, this application provides an intelligent detection method and device for deformation of steel wire rope in cages based on image recognition.

[0005] In a first aspect, this application provides an intelligent detection method for deformation of steel wire rope in cages based on image recognition, the method comprising:

[0006] Based on the cage start-stop status and recent operating parameters, the acquisition frequency is obtained, and the rope diameter image sequence and twist pitch image sequence are acquired. Defect identification is performed based on the rope diameter image sequence and the twist image sequence to obtain rope diameter defect parameters and twist defect parameters. By combining the rope diameter image sequence and the twist length image sequence, surface anomaly identification is performed to obtain surface anomaly parameters; Based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating conditions, the deformation of the cage wire rope is determined, the cage wire rope deformation identification result is obtained, and based on the cage wire rope deformation identification result, the acquisition frequency is adjusted to perform intelligent detection of cage wire rope deformation in the next identification cycle.

[0007] Secondly, this application provides an intelligent detection device for deformation of wire rope in cages based on image recognition, including: The image acquisition module is used to acquire the acquisition frequency and acquire the rope diameter image sequence and twist pitch image sequence based on the cage start-up and shutdown status and recent operating parameters. The defect identification module is used to identify defects based on the rope diameter image sequence and the twist length image sequence, and to obtain rope diameter defect parameters and twist length defect parameters. The surface anomaly identification module is used to integrate the rope diameter image sequence and the twist pitch image sequence to identify surface anomalies and obtain surface anomaly parameters. The deformation recognition module is used to determine the deformation of the cage wire rope based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating conditions parameters, to obtain the cage wire rope deformation recognition result, and to adjust the acquisition frequency based on the cage wire rope deformation recognition result to perform intelligent detection of cage wire rope deformation in the next recognition cycle.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application proposes an intelligent detection method and device for wire rope deformation in cages based on image recognition. By dynamically integrating the real-time start / stop status of the cage and historical operating parameters into the detection process, the image acquisition frequency is adaptively adjusted, thereby optimizing system resource allocation while ensuring that no key data is missed. The method comprehensively acquires two types of image sequences: rope diameter and lay length. Defect parameters are extracted separately using dedicated recognizers. Then, spatiotemporal registration and joint analysis of the two types of images are performed to identify surface anomalies. This multi-dimensional and collaborative image recognition mechanism overcomes the limitations of single-parameter detection and can more comprehensively characterize the structural and surface composite deformation of the wire rope. Furthermore, a confidence test based on recent operating conditions is introduced. In the verification stage, the reliability of the identified defect parameters is verified, and different abnormal patterns of the wire rope are identified by analyzing the temporal correlation between the rope diameter and lay pitch defect parameters. This ensures that the final deformation identification result not only includes the degree and location of the defect, but also reveals the potential type and cause of the defect. Compared with traditional fixed-frequency, single-dimensional detection methods lacking verification, the technical solution provided in this application significantly improves the intelligence level of the entire detection process. The adaptive acquisition mechanism ensures that the detection behavior matches the actual operating risks of the equipment, the multi-source information fusion and verification mechanism significantly enhances the accuracy and robustness of the detection results, and the ability to identify abnormal patterns provides a more accurate basis for predictive maintenance. This application achieves the technical effect of reliable and efficient detection of wire rope deformation in cages. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1This is a flowchart illustrating the intelligent detection method for deformation of steel wire rope in cages based on image recognition, provided in an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of the intelligent detection device for deformation of steel wire rope in cages based on image recognition, provided in an embodiment of this application.

[0012] The components represented by each number in the attached diagram are explained below: Image acquisition module 100, defect recognition module 200, surface anomaly recognition module 300, deformation recognition module 400. Detailed Implementation

[0013] This application provides an intelligent detection method and device for deformation of cage wire rope based on image recognition, which is used to address the technical problem of poor reliability and efficiency in detecting deformation of cage wire rope in the prior art.

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0015] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0016] Example 1, as Figure 1 As shown, this application provides an intelligent detection method for deformation of steel wire rope in cages based on image recognition, wherein the method includes: S10: Based on the cage start / stop status and recent operating parameters, acquire the acquisition frequency and acquire the rope diameter image sequence and twist pitch image sequence.

[0017] In the field of image detection of wire ropes in cages, existing technologies typically employ fixed-frequency or manual triggering methods for image acquisition, which has significant drawbacks. Fixed-frequency acquisition cannot adapt to the dynamic changes in the actual operation of the cage: when the cage is in low-risk conditions such as being stopped, lightly loaded, or operating at low speed, high-frequency acquisition will generate a large amount of redundant image data, resulting in unnecessary consumption of storage space and computing resources, and increasing the system's processing burden; conversely, when the cage experiences high-risk stages such as heavy load, high-speed operation, frequent start-stop, or historical data indicating recent abnormal conditions, fixed frequency may fail to capture the transient or accelerated deformation characteristics of the wire rope due to insufficient acquisition density.

[0018] Step S10 in the method provided in this application embodiment includes: Obtain the start-stop status of the cage and recent operating parameters, wherein the operating parameters include cage load parameters and hoisting speed parameters; Based on the cage start-stop status, a basic acquisition frequency is obtained, and based on the ratio of the operating parameters and historical high-frequency operating parameters, an acquisition frequency correction coefficient is obtained to correct the basic acquisition frequency, thereby obtaining the acquisition frequency. The steel wire rope of the cage is acquired using the acquisition frequency to obtain the rope diameter image sequence and the twist distance image sequence.

[0019] In this embodiment of the application, based on the start-stop status of the cage and recent operating parameters, the acquisition frequency is obtained, and the rope diameter image sequence and twist pitch image sequence are acquired.

[0020] Specifically, firstly, the start / stop status and recent operating parameters of the cage are acquired. These operating parameters include cage load parameters and hoisting speed parameters. For example, status sensors and operating condition sensors in the cage control system are used to acquire real-time data. The status sensors provide the start / stop status of the cage, i.e., whether it is currently operating and hoisting or stationary. The operating condition sensors include a weight sensor and a speed encoder, which provide the cage load parameters and hoisting speed parameters, respectively. For example, the sensors read that the current load is 5 tons and the current hoisting speed is 2 meters per second.

[0021] Further, based on the cage's start / stop status, a basic acquisition frequency is obtained, and based on the ratio of the operating parameters to historical high-frequency operating parameters, an acquisition frequency correction coefficient is obtained to correct the basic acquisition frequency, thus obtaining the acquisition frequency. For example, based on empirical presets: when the cage is in operation, the basic acquisition frequency is set to 10 frames per second to ensure sufficient sampling under dynamic conditions; when the cage is stopped, the basic acquisition frequency is set to 1 frame per second to reduce redundant data. Further, historical high-frequency operating parameters are obtained. These parameters can be common peak values ​​from the cage's historical operation, for example, setting the historical high-frequency load to 10 tons and the historical high-frequency speed to 4 meters per second. The correction coefficient is obtained by the ratio of the current operating parameters to the historical high-frequency operating parameters. Specifically, a weighted average method is used for calculation: Acquisition frequency correction coefficient = (current load / historical high-frequency load + current lifting speed / historical high-frequency speed) / 2 = (0.5 + 0.5) / 2 = 0.5. The obtained correction coefficient reflects the intensity of the current operating condition relative to the historical high frequency. Based on the aforementioned acquisition frequency correction coefficient, the base acquisition frequency is corrected to obtain the final acquisition frequency. Acquisition frequency = base acquisition frequency × correction coefficient. For example, if the base acquisition frequency during cage operation is 10 frames per second and the correction coefficient is 0.5, then the acquisition frequency = 10 × 0.5 = 5 frames per second. This frequency will dynamically adapt to changes in operating conditions to achieve resource optimization.

[0022] Furthermore, the aforementioned acquisition frequency is used to acquire images of the cage wire rope, obtaining the rope diameter image sequence and the lay length image sequence. Further, two industrial cameras are used for image acquisition, one aligned with the wire rope's diameter direction and the other with the lay length direction. The cameras continuously capture images at the calculated acquisition frequency, acquiring digital image sequences. The rope diameter image sequence captures visual information about the wire rope's diameter direction, and the lay length image sequence captures visual information about the wire rope's lay length direction. Both sequences are stored separately as input data for subsequent defect identification.

[0023] By incorporating cage start-stop status and recent operating parameters as decision-making criteria, the image acquisition frequency is dynamically acquired, enabling targeted acquisition of key image sequences for rope diameter and twist pitch. This achieves optimal allocation of detection resources. By linking the acquisition frequency with actual operational risks, the sampling density is automatically increased under complex operating conditions to ensure data completeness, while the frequency is reasonably reduced under simple operating conditions to minimize redundancy, thereby significantly improving the overall efficiency of the system.

[0024] S20: Based on the rope diameter image sequence and the twist image sequence, perform defect identification to obtain rope diameter defect parameters and twist defect parameters.

[0025] For cage wire ropes, abnormal rope diameter and abnormal lay are two key indicators characterizing different mechanical damage modes. They may occur independently or be related. Existing techniques often analyze these two separately or focus on only one, leading to poor accuracy in identification results.

[0026] Step S20 in the method provided in this application embodiment includes: Construct a rope diameter defect identifier and a twist pitch parameter identifier; The construction of the rope diameter defect identifier and the twist pitch parameter identifier includes: Based on convolutional neural networks, a rope diameter defect identifier and a twist pitch parameter identifier are constructed. Obtain an image of a sample wire rope of the same specifications and model as the wire rope, and label and obtain the sample rope diameter defect parameter set and the sample lay distance defect parameter set, wherein the sample rope diameter defect parameter and the sample lay distance defect parameter both include the defect degree and the defect location. The sample wire rope image is scaled and segmented to obtain a sample rope diameter image sequence and a sample twist image sequence. The rope diameter defect identifier and the twist parameter identifier are trained until convergence by combining the sample rope diameter defect parameter set and the sample twist defect parameter set. The rope diameter image sequence and the twist length image sequence are respectively input into the rope diameter defect identifier and the twist length parameter identifier to identify and obtain the rope diameter defect parameters and the twist length defect parameters.

[0027] In this embodiment of the application, defect identification is performed based on the rope diameter image sequence and the twist length image sequence to obtain rope diameter defect parameters and twist length defect parameters.

[0028] Specifically, firstly, a rope diameter defect identifier and a twist pitch parameter identifier are constructed.

[0029] Based on convolutional neural networks, a rope diameter defect identifier and a twist parameter identifier are constructed. Exemplarily, the same structure is used to construct the system based on the convolutional neural network. The first layer is a convolutional layer with 64 3×3 filters using the ReLU activation function; the second layer is a convolutional layer with 128 3×3 filters, also using the ReLU activation function; the third layer is a convolutional layer with 256 3×3 filters, using the ReLU activation function. A 2×2 max-pooling layer is connected after each convolutional layer. Finally, the feature map is flattened and connected to a fully connected layer with two output nodes, used to predict the defect severity and defect location, respectively.

[0030] Further, images of sample wire ropes of the same specifications and model as the wire rope to be tested are acquired, and the sample rope diameter defect parameter set and sample lay distance defect parameter set are labeled. Both the sample rope diameter defect parameter and the sample lay distance defect parameter include the degree and location of the defect. For example, used wire rope samples of the same specifications and model as the wire rope to be tested are acquired, and multiple sample images containing various defects such as deformation are taken from the rope diameter direction and the lay distance direction using an industrial camera. The degree of defect in each sample image is labeled manually as a percentage, for example, 0.2 represents slight deformation, and the location of the defect is labeled using the position of the defect center point, for example, (5m, 30°). This labeled information constitutes the sample rope diameter defect parameter set and the sample lay distance defect parameter set.

[0031] Further, the sample wire rope images are scaled and segmented to obtain sample rope diameter image sequences and sample twist distance image sequences. These are then used to train the rope diameter defect recognizer and twist distance parameter recognizer until convergence, in conjunction with the sample rope diameter defect parameter set and the sample twist distance defect parameter set. For example, the original sample images are preprocessed by scaling all images to a fixed size of 256×256 pixels, and then sequentially segmented into multiple 64×64 pixel small image blocks, thus forming the sample rope diameter image sequence and the sample twist distance image sequence. Using the sample rope diameter image sequence, the sample twist distance image sequence, and their corresponding labeled parameter sets, a supervised learning method is used to train the rope diameter defect recognizer and the twist distance parameter recognizer respectively. During training, the image sequences are input into the model, and the model outputs the predicted defect degree and location. The mean squared error between this prediction and the true labeled value is calculated as the loss function. The Adam optimizer is used for backpropagation to update the model parameters. This process is iterated repeatedly until the model's loss on the validation set no longer decreases significantly, i.e., training has converged.

[0032] Further, the rope diameter image sequence and the twist length image sequence are input into the rope diameter defect recognizer and the twist length parameter recognizer, respectively, to identify and obtain the rope diameter defect parameters and twist length defect parameters. The acquired rope diameter image sequence is input into the pre-trained rope diameter defect recognizer. The model processes each frame of the sequence and automatically outputs the recognition result, i.e., the rope diameter defect parameters, including the degree and location of the rope diameter defect. Similarly, the acquired twist length image sequence is input into the pre-trained twist length parameter recognizer. The model also analyzes the sequence frame by frame and automatically outputs the recognition result, i.e., the twist length defect parameters, including the degree and location of the twist length defect.

[0033] By constructing and applying a specialized identifier, the automatic defect identification and parameter extraction of the rope diameter image sequence and the twist length image sequence are performed respectively, ensuring the comprehensiveness of the analysis. It can simultaneously capture the two main types of geometric deformation that may exist in the wire rope, avoiding misjudgment of the overall state that may be caused by only detecting a single indicator.

[0034] S30: Combine the rope diameter image sequence and the twist length image sequence to perform surface anomaly identification and obtain surface anomaly parameters.

[0035] Surface anomalies in wire ropes are a significant precursor to their failure, but accurately identifying these anomalies is particularly difficult. Relying solely on images from a single viewpoint for surface detection presents challenges due to the limitation of the viewpoint: certain surface defects may be occluded, incompletely displayed, or confused with the background texture at a particular viewpoint, leading to a decrease in recognition rate or insufficient feature extraction.

[0036] Step S30 in the method provided in this application embodiment includes: Spatiotemporal registration is performed on the rope diameter image sequence and the twist pitch image sequence to obtain a composite image pair; The rope diameter image sequence and the twist length image sequence are respectively subjected to surface anomaly identification, and the surface anomaly identification results of each integrated image pair are obtained. Consistency verification is performed, and high consistency surface identification results are obtained as surface anomaly parameters. The surface anomaly parameters include the degree of anomaly and the location of the anomaly.

[0037] In this embodiment, surface anomaly identification is performed by combining the rope diameter image sequence and the twist image sequence to obtain surface anomaly parameters. For example, based on the timestamp information synchronously recorded by the two cameras during acquisition, a rope diameter image and a twist image acquired at the same time are paired together to form a composite image pair. For instance, rope diameter image A acquired at timestamp "12:00:00.5" is paired with twist image B acquired at the same time to obtain a composite image pair (A, B). This process is repeated for all image frames with matching timestamps to obtain a series of composite image pairs arranged chronologically.

[0038] Further, surface anomaly identification is performed on the rope diameter image sequence and the twist length image sequence respectively, and the surface anomaly identification results of each integrated image pair are obtained. Consistency verification is performed, and high-consistency surface identification results are obtained as surface anomaly parameters, wherein the surface anomaly parameters include anomaly degree and anomaly location. For example, a surface anomaly identification model is constructed. This model adopts a four-layer convolutional neural network structure, implemented using the TensorFlow framework. The first layer is a convolutional layer using 32 3×3 filters; the second layer is a convolutional layer using 64 3×3 filters; the third layer is a convolutional layer using 128 3×3 filters; each convolutional layer is followed by a 2×2 max pooling layer and a ReLU activation function. Finally, a fully connected layer is connected to output the anomaly degree and anomaly location. The model is trained using historical wire rope surface images (including annotations such as broken wires and rust) until convergence. Then, each frame of the rope diameter image sequence is input into this trained model, and the model outputs the surface anomaly identification result for that frame, for example, anomaly degree 0.3, location (8m, 18°). Similarly, each frame in the twist image sequence is input into the same model to obtain another set of recognition results. Further, consistency checks are performed to obtain highly consistent surface anomaly parameters. For each composite image pair, the recognition results for the rope diameter image and the twist image are extracted. The consistency difference between these two results is calculated: Difference value = |Anomaly degree of the rope diameter image - Anomaly degree of the twist image|. An empirical consistency threshold is set, for example, 0.1. If the calculated difference value is less than this threshold, the recognition results from the two perspectives are considered to have high consistency. In this case, the average of the two recognition results is taken as the final surface anomaly parameter, including the averaged anomaly degree and the averaged anomaly location coordinates. If the difference value is greater than the threshold, the current recognition result of that composite image pair is discarded and considered low-consistency noise. After traversing all composite image pairs, all the remaining highly consistent recognition results constitute the final set of obtained surface anomaly parameters.

[0039] By integrating image sequences of rope diameter and twist pitch for surface anomaly identification, and performing spatiotemporal registration and comprehensive analysis on image sequences from different angles, anomalous features that are blurry, hidden, or easily disturbed from a single perspective can be more clearly captured and confirmed with the confirmation or supplementation from another perspective.

[0040] S40: Based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating condition parameters, determine the deformation of the cage wire rope, obtain the cage wire rope deformation identification result, and based on the cage wire rope deformation identification result, adjust the acquisition frequency to perform intelligent detection of cage wire rope deformation in the next identification cycle.

[0041] After obtaining the rope diameter, twist pitch defect parameters, and surface anomaly parameters respectively, the key to realizing the detection value lies in how to effectively integrate these multi-source and heterogeneous detection information.

[0042] Step S40 in the method provided in this application embodiment includes: Based on recent operating conditions, the confidence levels of the rope diameter defect parameters and the twist pitch defect parameters are verified to obtain the confidence levels of the rope diameter defect and the twist pitch defect. When the confidence levels of the rope diameter defect and the lay length defect are both greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained. Specifically, when both the confidence levels of the rope diameter defect and the lay length defect are greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained, including: Based on the rope diameter defect parameters and the lay pitch defect parameters, the abnormal mode of the cage wire rope is obtained. The abnormal mode of the cage wire rope is obtained based on the rope diameter defect parameter and the lay length defect parameter, including: Calculate the time-series correlation coefficient between the rope diameter defect parameter and the twist pitch defect parameter, wherein the time-series correlation coefficient is the statistical correlation between the rope diameter defect parameter and the twist pitch defect parameter; When the time-series correlation coefficient is positive, the lifting depth parameter in the recent operating condition parameters is used to determine whether it is a local loosening-type coordinated deformation. When the time-series correlation coefficient is negative, the lifting load parameter in the recent operating condition parameters is used to determine whether it is a local wear-related contradictory deformation. By combining the location information in the rope diameter defect parameters, the lay pitch defect parameters, and the surface anomaly parameters, as well as the cage wire rope anomaly pattern, the deformation identification result of the cage wire rope is obtained. If either the confidence level of the rope diameter defect or the confidence level of the twist pitch defect is less than the confidence threshold, the corresponding image sequence is re-acquired and anomaly identification is performed, and the confidence level is verified again. If either one is still less than the confidence threshold, an anomaly warning is triggered. Based on the abnormal patterns of the cage wire rope in the cage wire rope deformation identification results, the periodic acquisition frequency is matched and obtained. A confidence correction coefficient is obtained by weighting the confidence levels of the rope diameter defect and the twist pitch defect. The confidence correction coefficient is used to correct the periodic acquisition frequency, and the corrected acquisition frequency is obtained as the base acquisition frequency for the next identification period.

[0043] In this embodiment, based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating condition parameters, the deformation of the cage wire rope is determined, the cage wire rope deformation identification result is obtained, and based on the cage wire rope deformation identification result, the acquisition frequency is adjusted to perform intelligent detection of cage wire rope deformation in the next identification cycle.

[0044] Specifically, firstly, based on recent working conditions, the confidence levels of the rope diameter defect parameters and the lay length defect parameters are verified to obtain the confidence levels of the rope diameter defects and lay length defects. For example, a working condition intensity index is calculated to characterize the severity of recent working conditions: Working Condition Intensity Index = (Recent Average Load / Rated Load + Recent Average Lifting Speed ​​ / Rated Lifting Speed) / 2. For instance, if the recent average load is 6 tons and the rated load is 10 tons; the recent average lifting speed is 3 m / s and the rated lifting speed is 5 m / s, then the Working Condition Intensity Index = (6 / 10 + 3 / 5) / 2 = 0.6. Next, defect severity values ​​are extracted from the rope diameter defect parameter sequence, and their rate of change is calculated. For example, if the sequence is [0.1, 0.12, 0.15], the rate of change = (0.15 - 0.1) ÷ 0.1 = 0.5. Therefore, the confidence level of the rope diameter defect = 1 - |Rate of Change - Working Condition Intensity Index| = 1 - |0.5 - 0.6| = 0.9. The same method was used to calculate and obtain the confidence level of the twist defect.

[0045] Furthermore, when the confidence levels of the rope diameter defect and the lay length defect are both greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained.

[0046] First, based on the rope diameter defect parameters and the lay length defect parameters, the abnormal mode of the cage wire rope is obtained.

[0047] Specifically, the time-series correlation coefficient between the rope diameter defect severity sequence and the twist pitch defect severity sequence is calculated using the Pearson correlation coefficient formula: Correlation coefficient = Covariance(rope diameter sequence, twist pitch sequence) ÷ [Standard deviation(rope diameter sequence) × Standard deviation(twist pitch sequence)]. For example, a correlation coefficient of 0.98 is obtained, which is positive.

[0048] Furthermore, when the time-series correlation coefficient is positive, the lifting depth parameter in the recent operating condition parameters is used to determine whether it is a localized loosening-type coordinated deformation. For example, the determination is made by combining the lifting depth parameter in the recent operating condition parameters. If the lifting depth is 500 meters, since the correlation coefficient is positive and the lifting depth is relatively large, it is determined to be a "localized loosening-type coordinated deformation".

[0049] Furthermore, when the time-series correlation coefficient is negative, the lifting load parameter in the recent operating condition parameters is used to determine whether it is a localized wear-related contradictory deformation. For example, if the correlation coefficient is negative, the lifting load parameter is used for judgment. If the load is large and the time-series correlation coefficient is negative, it is determined to be a localized wear-related contradictory deformation.

[0050] Furthermore, by combining the location information in the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, along with the anomaly pattern of the cage wire rope, the deformation identification result of the cage wire rope is obtained. For example, by combining the location information in the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, the Euclidean distance between these locations is calculated. If the distance is less than a set threshold, such as 0.1m, the locations are considered close. The deformation identification result of the cage wire rope is obtained by comprehensively considering the anomaly pattern. For example, there may be localized loose strand-like coordinated deformation and surface defects, located near the coordinates (150m, 30°).

[0051] Furthermore, when either the confidence level of the rope diameter defect or the confidence level of the twist pitch defect is less than a confidence threshold, the corresponding image sequence is re-acquired and anomaly identification is performed, and the confidence level is verified again. If either confidence level is still less than the confidence threshold, an anomaly warning is triggered. For example, when either confidence level is less than the confidence threshold, the corresponding industrial camera is controlled to re-acquire the rope diameter or twist pitch image sequence, and defect identification is re-performed to obtain new defect parameters. The confidence level is verified again. If the confidence level is still less than the threshold, an anomaly warning signal is sent through the system network, indicating a possible anomaly or a possible anomaly in the industrial camera.

[0052] Furthermore, based on the abnormal patterns of the cage wire rope in the deformation identification results, a periodic acquisition frequency is matched and obtained. For example, based on the abnormal patterns in the deformation identification results, a periodic acquisition frequency is matched from a preset mapping table, such as "local loose strand cooperative deformation" corresponding to 10 frames per second, and "local wear contradictory deformation" corresponding to 15 frames per second.

[0053] Furthermore, a confidence correction coefficient is obtained by weighting the confidence levels of the rope diameter defect and the twist pitch defect. For example, the confidence correction coefficient = rope diameter defect confidence level × rope diameter weight + twist pitch defect confidence level × twist pitch weight. For instance, if the rope diameter defect confidence level is 0.9 and the twist pitch defect confidence level is 0.85, and both weights are 0.5, then the confidence correction coefficient = 0.9 × 0.5 + 0.85 × 0.5 = 0.875.

[0054] Furthermore, the confidence correction coefficient is used to correct the periodic acquisition frequency to obtain the corrected acquisition frequency, which serves as the base acquisition frequency for the next identification period. For example, the corrected acquisition frequency = periodic acquisition frequency × confidence correction coefficient, such as 10 × 0.875 = 8.75 frames per second, rounded down to 9 frames per second as the base acquisition frequency for the next identification period.

[0055] Intelligent detection of wire rope deformation in cages was achieved through multi-parameter fusion diagnosis and closed-loop feedback adjustment. By combining three types of parameters—rope diameter, lay pitch, and surface area—with real-time and historical operating condition parameters for comprehensive analysis, it can not only determine the existence of defects but also preliminarily assess the defect patterns. The output of the cage wire rope deformation identification result is a more informative and decision-supporting comprehensive diagnostic conclusion. Furthermore, based on the results of this identification, the image acquisition frequency for the next cycle can be intelligently adjusted. When a potentially high-risk deformation pattern is identified, the subsequent monitoring frequency is automatically increased for close tracking; when the condition is good, the frequency is appropriately reduced to save resources and improve efficiency.

[0056] Example 2, as Figure 2 As shown, based on the same inventive concept as the image recognition-based intelligent detection method for cage wire rope deformation provided in Embodiment 1, this embodiment of the invention also provides an image recognition-based intelligent detection device for cage wire rope deformation, comprising: The image acquisition module 100 is used to acquire the acquisition frequency and acquire the rope diameter image sequence and twist pitch image sequence based on the cage start-up and shutdown status and recent working condition parameters. The defect identification module 200 is used to identify defects based on the rope diameter image sequence and the twist length image sequence, and to obtain rope diameter defect parameters and twist length defect parameters. The surface anomaly identification module 300 is used to integrate the rope diameter image sequence and the twist pitch image sequence to identify surface anomalies and obtain surface anomaly parameters. The deformation recognition module 400 is used to determine the deformation of the cage wire rope based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating condition parameters, to obtain the cage wire rope deformation recognition result, and to adjust the acquisition frequency based on the cage wire rope deformation recognition result to perform intelligent detection of cage wire rope deformation in the next recognition cycle.

[0057] In one embodiment, the image acquisition module 100 is further configured to: Obtain the start-stop status of the cage and recent operating parameters, wherein the operating parameters include cage load parameters and hoisting speed parameters; Based on the cage start-stop status, a basic acquisition frequency is obtained, and based on the ratio of the operating parameters and historical high-frequency operating parameters, an acquisition frequency correction coefficient is obtained to correct the basic acquisition frequency, thereby obtaining the acquisition frequency. The steel wire rope of the cage is acquired using the acquisition frequency to obtain the rope diameter image sequence and the twist distance image sequence.

[0058] In one embodiment, the defect identification module 200 is further configured to: Construct a rope diameter defect identifier and a twist pitch parameter identifier; The construction of the rope diameter defect identifier and the twist pitch parameter identifier includes: Based on convolutional neural networks, a rope diameter defect identifier and a twist pitch parameter identifier are constructed. Obtain an image of a sample wire rope of the same specifications and model as the wire rope, and label and obtain the sample rope diameter defect parameter set and the sample lay distance defect parameter set, wherein the sample rope diameter defect parameter and the sample lay distance defect parameter both include the defect degree and the defect location. The sample wire rope image is scaled and segmented to obtain a sample rope diameter image sequence and a sample twist image sequence. The rope diameter defect identifier and the twist parameter identifier are trained until convergence by combining the sample rope diameter defect parameter set and the sample twist defect parameter set. The rope diameter image sequence and the twist length image sequence are respectively input into the rope diameter defect identifier and the twist length parameter identifier to identify and obtain the rope diameter defect parameters and the twist length defect parameters.

[0059] In one embodiment, the surface anomaly identification module 300 is further configured to: Spatiotemporal registration is performed on the rope diameter image sequence and the twist pitch image sequence to obtain a composite image pair; The rope diameter image sequence and the twist length image sequence are respectively subjected to surface anomaly identification, and the surface anomaly identification results of each integrated image pair are obtained. Consistency verification is performed, and high consistency surface identification results are obtained as surface anomaly parameters. The surface anomaly parameters include the degree of anomaly and the location of the anomaly.

[0060] In one embodiment, the deformation recognition module 400 is further configured to: Based on recent operating conditions, the confidence levels of the rope diameter defect parameters and the twist pitch defect parameters are verified to obtain the confidence levels of the rope diameter defect and the twist pitch defect. When the confidence levels of the rope diameter defect and the lay length defect are both greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained. Specifically, when both the confidence levels of the rope diameter defect and the lay length defect are greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained, including: Based on the rope diameter defect parameters and the lay pitch defect parameters, the abnormal mode of the cage wire rope is obtained. The abnormal mode of the cage wire rope is obtained based on the rope diameter defect parameter and the lay length defect parameter, including: Calculate the time-series correlation coefficient between the rope diameter defect parameter and the twist pitch defect parameter, wherein the time-series correlation coefficient is the statistical correlation between the rope diameter defect parameter and the twist pitch defect parameter; When the time-series correlation coefficient is positive, the lifting depth parameter in the recent operating condition parameters is used to determine whether it is a local loosening-type coordinated deformation. When the time-series correlation coefficient is negative, the lifting load parameter in the recent operating condition parameters is used to determine whether it is a local wear-related contradictory deformation. By combining the location information in the rope diameter defect parameters, the lay pitch defect parameters, and the surface anomaly parameters, as well as the cage wire rope anomaly pattern, the deformation identification result of the cage wire rope is obtained. If either the confidence level of the rope diameter defect or the confidence level of the twist pitch defect is less than the confidence threshold, the corresponding image sequence is re-acquired and anomaly identification is performed, and the confidence level is verified again. If either one is still less than the confidence threshold, an anomaly warning is triggered. Based on the abnormal patterns of the cage wire rope in the cage wire rope deformation identification results, the periodic acquisition frequency is matched and obtained. A confidence correction coefficient is obtained by weighting the confidence levels of the rope diameter defect and the twist pitch defect. The confidence correction coefficient is used to correct the periodic acquisition frequency, and the corrected acquisition frequency is obtained as the base acquisition frequency for the next identification period.

[0061] In summary, the embodiments of this application have at least the following technical effects: This application proposes an intelligent detection method and device for wire rope deformation in cages based on image recognition. By dynamically integrating the real-time start / stop status of the cage and historical operating parameters into the detection process, the image acquisition frequency is adaptively adjusted, thereby optimizing system resource allocation while ensuring that no key data is missed. The method comprehensively acquires two types of image sequences: rope diameter and lay length. Defect parameters are extracted separately using dedicated recognizers. Then, spatiotemporal registration and joint analysis of the two types of images are performed to identify surface anomalies. This multi-dimensional and collaborative image recognition mechanism overcomes the limitations of single-parameter detection and can more comprehensively characterize the structural and surface composite deformation of the wire rope. Furthermore, a confidence test based on recent operating conditions is introduced. In the verification stage, the reliability of the identified defect parameters is verified, and different abnormal patterns of the wire rope are identified by analyzing the temporal correlation between the rope diameter and lay pitch defect parameters. This ensures that the final deformation identification result not only includes the degree and location of the defect, but also reveals the potential type and cause of the defect. Compared with traditional fixed-frequency, single-dimensional detection methods lacking verification, the technical solution provided in this application significantly improves the intelligence level of the entire detection process. The adaptive acquisition mechanism ensures that the detection behavior matches the actual operating risks of the equipment, the multi-source information fusion and verification mechanism significantly enhances the accuracy and robustness of the detection results, and the ability to identify abnormal patterns provides a more accurate basis for predictive maintenance. This application achieves the technical effect of reliable and efficient detection of wire rope deformation in cages.

[0062] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0063] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0064] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An intelligent detection method for deformation of steel wire rope in cages based on image recognition, characterized in that, include: Based on the cage start-stop status and recent operating parameters, the acquisition frequency is obtained, and the rope diameter image sequence and twist pitch image sequence are acquired. Defect identification is performed based on the rope diameter image sequence and the twist image sequence to obtain rope diameter defect parameters and twist defect parameters. By combining the rope diameter image sequence and the twist length image sequence, surface anomaly identification is performed to obtain surface anomaly parameters; Based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating conditions, the deformation of the cage wire rope is determined, the cage wire rope deformation identification result is obtained, and based on the cage wire rope deformation identification result, the acquisition frequency is adjusted to perform intelligent detection of cage wire rope deformation in the next identification cycle.

2. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 1, characterized in that, Based on the cage's start / stop status and recent operating parameters, the acquisition frequency is determined, and image sequences of rope diameter and twist pitch are acquired, including: Obtain the start-stop status of the cage and recent operating parameters, wherein the operating parameters include cage load parameters and hoisting speed parameters; Based on the cage start-stop status, a basic acquisition frequency is obtained, and based on the ratio of the operating parameters and historical high-frequency operating parameters, an acquisition frequency correction coefficient is obtained to correct the basic acquisition frequency, thereby obtaining the acquisition frequency. The steel wire rope of the cage is acquired using the acquisition frequency to obtain the rope diameter image sequence and the twist distance image sequence.

3. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 1, characterized in that, Defect identification is performed based on the rope diameter image sequence and the twist length image sequence to obtain rope diameter defect parameters and twist length defect parameters, including: Construct a rope diameter defect identifier and a twist pitch parameter identifier; The rope diameter image sequence and the twist length image sequence are respectively input into the rope diameter defect identifier and the twist length parameter identifier to identify and obtain the rope diameter defect parameters and the twist length defect parameters.

4. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 3, characterized in that, The construction of the rope diameter defect identifier and the twist pitch parameter identifier includes: Based on convolutional neural networks, a rope diameter defect identifier and a twist pitch parameter identifier are constructed. Obtain an image of a sample wire rope of the same specifications and model as the wire rope, and label and obtain the sample rope diameter defect parameter set and the sample lay distance defect parameter set, wherein the sample rope diameter defect parameter and the sample lay distance defect parameter both include the defect degree and the defect location. The sample wire rope images are scaled and segmented to obtain sample rope diameter image sequences and sample twist distance image sequences. Combined with the sample rope diameter defect parameter set and sample twist distance defect parameter set, the rope diameter defect recognizer and twist distance parameter recognizer are trained until convergence.

5. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 1, characterized in that, By combining the rope diameter image sequence and the twist length image sequence, surface anomaly identification is performed to obtain surface anomaly parameters, including: Spatiotemporal registration is performed on the rope diameter image sequence and the twist pitch image sequence to obtain a composite image pair; The rope diameter image sequence and the twist length image sequence are respectively subjected to surface anomaly identification, and the surface anomaly identification results of each integrated image pair are obtained. Consistency verification is performed, and high consistency surface identification results are obtained as surface anomaly parameters. The surface anomaly parameters include the degree of anomaly and the location of the anomaly.

6. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 1, characterized in that, Based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating condition parameters, the deformation of the cage wire rope is determined, and the cage wire rope deformation identification result is obtained, including: Based on recent operating conditions, the confidence levels of the rope diameter defect parameters and the twist pitch defect parameters are verified to obtain the confidence levels of the rope diameter defect and the twist pitch defect. When the confidence levels of the rope diameter defect and the lay length defect are both greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained. If either the confidence level of the rope diameter defect or the confidence level of the twist pitch defect is less than the confidence threshold, the corresponding image sequence is re-acquired and anomaly identification is performed, and the confidence level is verified again. If either one is still less than the confidence threshold, an anomaly warning is triggered.

7. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 6, characterized in that, When both the confidence level of the rope diameter defect and the confidence level of the lay length defect are greater than the confidence threshold, the deformation of the cage wire rope is identified by combining the surface anomaly parameters, and the deformation identification result of the cage wire rope is obtained, including: Based on the rope diameter defect parameters and the lay pitch defect parameters, the abnormal mode of the cage wire rope is obtained. By combining the location information in the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, as well as the abnormal pattern of the cage wire rope, the deformation identification result of the cage wire rope is obtained.

8. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 7, characterized in that, Based on the rope diameter defect parameters and the lay length defect parameters, the abnormal modes of the cage wire rope are obtained, including: Calculate the time-series correlation coefficient between the rope diameter defect parameter and the twist pitch defect parameter, wherein the time-series correlation coefficient is the statistical correlation between the rope diameter defect parameter and the twist pitch defect parameter; When the time-series correlation coefficient is positive, the lifting depth parameter in the recent operating condition parameters is used to determine whether it is a local loosening-type coordinated deformation. When the time-series correlation coefficient is negative, the lifting load parameter in the recent operating condition parameters is used to determine whether it is a local wear-related contradictory deformation.

9. The intelligent detection method for deformation of steel wire rope in cages based on image recognition according to claim 1, characterized in that, Based on the deformation identification results of the cage wire rope, the acquisition frequency is adjusted to perform intelligent detection of cage wire rope deformation in the next identification cycle, including: Based on the abnormal patterns of the cage wire rope in the cage wire rope deformation identification results, the periodic acquisition frequency is matched and obtained. A confidence correction coefficient is obtained by weighting the confidence levels of the rope diameter defect and the twist pitch defect. The confidence correction coefficient is used to correct the periodic acquisition frequency, and the corrected acquisition frequency is obtained as the base acquisition frequency for the next identification period.

10. An intelligent detection device for deformation of steel wire rope in cages based on image recognition, characterized in that, For implementing the image recognition-based intelligent detection method for deformation of steel wire rope in cages according to any one of claims 1-9, the apparatus comprises: The image acquisition module is used to acquire the acquisition frequency and acquire the rope diameter image sequence and twist pitch image sequence based on the cage start-up and shutdown status and recent operating parameters. The defect identification module is used to identify defects based on the rope diameter image sequence and the twist length image sequence, and to obtain rope diameter defect parameters and twist length defect parameters. The surface anomaly identification module is used to integrate the rope diameter image sequence and the twist pitch image sequence to identify surface anomalies and obtain surface anomaly parameters. The deformation recognition module is used to determine the deformation of the cage wire rope based on the rope diameter defect parameters, the lay length defect parameters, and the surface anomaly parameters, combined with the cage start-up and shutdown status and recent operating conditions parameters, to obtain the cage wire rope deformation recognition result, and to adjust the acquisition frequency based on the cage wire rope deformation recognition result to perform intelligent detection of cage wire rope deformation in the next recognition cycle.

Images