A copper wire large-drawing machine anti-hanging detection method and system based on light curtain recognition

Abstract

The present application relates to the technical field of safety detection, and more particularly to a copper wire large-drawing machine anti-hanging detection method and system based on light curtain recognition, which comprises the following steps: for the risk area of the copper wire large-drawing machine pay-off rack, the area calibration and initialization of the safety light curtain are performed, and the detection reference is established; the on-off state of each light beam is collected in real time through the safety light curtain, the original signal is pre-processed to eliminate interference, and a time-varying shielding vector sequence is generated; the waveform signal of the number of continuous shielding light beams varying with time is extracted; the waveform signal is compared with at least one pre-stored human intrusion standard waveform template through dynamic time warping, and the waveform similarity is calculated; the waveform similarity is compared with the preset threshold, and the corresponding anti-hanging detection result is output. Through the present application, the core contradiction of false triggering and missed detection of the existing copper wire large-drawing machine anti-hanging detection is effectively solved, the human intrusion risk and the on-site interference can be accurately distinguished, the production continuity and the protection safety are taken into account, and the detection reliability is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of safety detection technology, and in particular to a method and system for detecting the anti-slip function of a large copper wire pulling machine based on light curtain recognition. Background Technology

[0002] The copper wire drawing machine is a core piece of equipment in metal wire processing. During the wire laying process, the copper rod runs at high speed under tension. If operators are accidentally entangled or hooked by the copper rod, they can easily be lifted or pulled into the equipment, causing serious personal injury accidents. Currently, for anti-slip protection in the wire laying area, the industry mostly uses mechanical limit switches, counterweight limiters, or manual video monitoring. Among these, mechanical limit switches are prone to malfunction due to oil contamination and vibration, and have a delayed response; counterweight limiters can only detect vertical displacement and cannot identify the risk of personnel entering from the side or bottom; manual monitoring suffers from visual fatigue and untimely response, and none of these methods can achieve reliable anti-slip protection.

[0003] Safety light curtains, as non-contact photoelectric detection devices, have been widely used in industrial intrusion prevention. However, existing intrusion detection methods based on light curtains cannot accurately distinguish between the effective risk of human intrusion and various interference signals in the complex operating environment of copper wire drawing machines. The contradiction between false triggering and missed detection cannot be reconciled, making it impossible to achieve reliable anti-crane protection. At the same time, existing methods rely solely on a fixed threshold for the number of beams blocked in a single frame, which cannot identify the essential attributes of the blocking event. This results in either frequent false triggering affecting production or a threshold that is too lenient, leading to missed detections and safety accidents. It is impossible to balance production efficiency and safety protection. Summary of the Invention

[0004] This invention provides a method for detecting the anti-slip of large copper wire drawing machines based on light curtain recognition, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for detecting the anti-slip function of a large copper wire drawing machine based on light curtain recognition, the method comprising: For the risk area of ​​the copper wire drawing machine's pay-off frame, the area of ​​the safety light curtain is calibrated and initialized to establish a detection benchmark; The on / off status of each beam is collected in real time by a safety light curtain, the original signal is preprocessed to remove interference, and a time-varying occlusion vector sequence is generated. Extract the waveform signal of the number of continuously blocked beams changing over time from the blockage vector sequence; The waveform signal is dynamically time-normalized and compared with at least one pre-stored standard waveform template for human intrusion to calculate the waveform similarity. The waveform similarity is compared with a preset threshold, and the corresponding anti-hanging detection result is output.

[0006] Furthermore, the vertical detection range of the safety light curtain extends from the ground to the upper limit of the dangerous height at which a person can be suspended.

[0007] Furthermore, the safety light curtain is a through-beam light curtain, with the transmitter and receiver of the safety light curtain respectively installed on both sides of the cable laying frame, and the detection plane of the safety light curtain is perpendicular to the ground.

[0008] Furthermore, the on / off status of each beam is acquired in real time through a safety light curtain, and the original signal is preprocessed to remove interference, generating a time-varying occlusion vector sequence, including: The received light intensity of each beam is scanned at a fixed sampling period, and those below a preset background threshold are judged as occlusion states. Median filtering is applied to the occlusion states of multiple consecutive frames, and the occlusion states of each frame after filtering are arranged into binary vectors according to the beam order, and stored in chronological order as the occlusion vector sequence.

[0009] Further, extracting the waveform signal of the number of continuously obstructed beams changing over time from the obstruction vector sequence includes: For each frame of occlusion vector, traverse the occlusion state of each beam and calculate the number of the longest consecutively occluded beams as the occlusion quantity feature value of that frame. Arrange the occlusion quantity feature values ​​of each frame into a one-dimensional sequence according to time order to obtain the waveform signal that changes with time.

[0010] Further, the waveform signal is dynamically time-warped and compared with at least one pre-stored standard waveform template for human intrusion to calculate waveform similarity, including: Determine the time axis correspondence between the real-time waveform signal and the standard waveform template, and set the search window for dynamic time warping; Calculate the distance between each point in the two sequences point by point, and construct the distance matrix using the cumulative distance recursive formula; The optimal alignment path is searched from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; The minimum cumulative distance is converted into a similarity score with a fixed value range through normalization.

[0011] Furthermore, the human intrusion standard waveform template includes at least three templates corresponding to different intrusion speeds, which are used to match the intrusion waveforms of personnel entering at slow speed, normal speed, and fast speed into the light curtain area, respectively.

[0012] Furthermore, the waveform similarity is compared with a preset threshold, and the corresponding anti-hanging detection result is output, including: The calculated waveform similarity is compared with a preset similarity threshold. If the waveform similarity reaches or exceeds the similarity threshold, it is determined that the current occlusion event matches the human intrusion standard waveform template, and a detection result indicating a risk of shoving is output. If the waveform similarity is lower than the similarity threshold, a detection result indicating no risk of suspension is output.

[0013] A copper wire drawing machine anti-slip detection system based on light curtain recognition, the system comprising: The detection baseline establishment module performs area calibration and initialization of the safety light curtain for the risk area of ​​the copper wire drawing machine's pay-off frame, and establishes the detection benchmark. The vector sequence generation module collects the on / off status of each beam in real time through the safety light curtain, preprocesses the original signal to remove interference, and generates an occlusion vector sequence that changes over time. The waveform signal extraction module extracts the waveform signal of the number of continuously blocked beams changing over time from the blockage vector sequence; The similarity calculation module performs dynamic time-normalization comparison between the waveform signal and at least one pre-stored human intrusion standard waveform template to calculate the waveform similarity. The detection result generation module compares the waveform similarity with a preset threshold and outputs the corresponding anti-hanging detection result.

[0014] Furthermore, the similarity calculation module includes: The search window setting unit determines the time axis correspondence between the real-time waveform signal and the standard waveform template, and sets the search window for dynamic time warping. The distance matrix construction unit calculates the distance between each point in the two sequences point by point, and constructs the distance matrix using the cumulative distance recursive formula; The optimal path search unit searches for the optimal alignment path from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; The similarity normalization unit converts the minimum cumulative distance into a similarity score with a fixed value range through normalization processing.

[0015] The technical solution of this invention can achieve the following technical effects: This solution effectively addresses the core pain points of existing anti-lift detection methods for large copper wire drawing machines, which cannot distinguish between human intrusion risks and on-site interference, and the irreconcilable contradiction between false triggers and missed detections. This solution abandons the traditional judgment logic of fixed thresholds for single frames. By extracting the time-series waveform signal of the number of consecutive beam obstructions during a person's intrusion, it fully preserves the dynamic characteristics of the intrusion action. Combined with a dynamic time warping algorithm and a standard template for human intrusion, it performs similarity matching, accurately distinguishing between effective risks and instantaneous interference from the essence of the action. This simultaneously achieves a significant reduction in both missed detection and false trigger rates, ensuring the inherent safety of on-site personnel while also ensuring the continuous and stable operation of the production line. It perfectly adapts to the anti-lift detection needs of complex operating scenarios for large copper wire drawing machines.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for detecting the anti-slip function of a large copper wire drawing machine based on light curtain recognition. Figure 2 A flowchart illustrating the process of generating a time-varying sequence of occlusion vectors; Figure 3 This is a flowchart illustrating the process of extracting the waveform signal of the number of continuously obstructed beams changing over time from an obstruction vector sequence. Figure 4 This is a flowchart illustrating the process of calculating waveform similarity. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1: like Figure 1 As shown, this application provides a method for detecting the anti-slip function of a large copper wire drawing machine based on light curtain recognition. The method includes: S1: For the risk area of ​​the copper wire drawing machine's pay-off frame, perform area calibration and initialization of the safety light curtain to establish a detection benchmark; S2: The on / off status of each beam is collected in real time through the safety light curtain, the original signal is preprocessed to remove interference, and a time-varying occlusion vector sequence is generated. S3: Extract the waveform signal of the number of continuously blocked beams changing over time from the blocking vector sequence; S4: Perform dynamic time warping comparison between the waveform signal and at least one pre-stored standard waveform template for human intrusion, and calculate the waveform similarity. S5: Compare the waveform similarity with the preset threshold and output the corresponding anti-hanging detection result.

[0022] Firstly, targeting the core risk area during copper wire drawing machine operations—where personnel are prone to entanglement or snagging on copper rods, potentially leading to hoisting accidents—the protection area of ​​the safety light curtain was calibrated and the system initialized, establishing a standard testing benchmark adapted to the on-site working environment. During the on-site implementation, priority was given to defining the core protection area where personnel would be directly exposed to the hoisting risk if they entered, based on the physical structure of the wire drawing machine, the activity range of personnel during normal operations, and the running path of the copper rods. The corresponding deployment location and protection boundary of the safety light curtain were then calibrated. After the system was powered on, priority was given to completing the communication synchronization calibration at both ends of the light curtain and the acquisition of reference signals under unobstructed conditions. This established a stable and unified standard testing benchmark, providing a consistent judgment basis for subsequent full-process testing.

[0023] This invention effectively solves the core contradiction of false triggering and missed detection in the existing anti-lifting detection of large copper wire drawing machines. It can accurately distinguish between the risk of human intrusion and on-site interference, taking into account both production continuity and protection safety, and greatly improving the reliability of detection.

[0024] By calibrating and initializing the safety light curtain, the on / off status of each beam within the protected area is collected in real time. The original on / off signals are preprocessed to remove various interference signals from the environment. The processed valid signals are then organized in chronological order to generate a continuously changing occlusion vector sequence. In the optimal implementation of actual operations, a fixed high-frequency sampling period is preferred to continuously scan the on / off status of all beams in the light curtain, ensuring that the entire process of occlusion events from occurrence to end can be captured completely. The preprocessing stage prioritizes removing instantaneous jump interference in the original signals that does not conform to the continuous change pattern, retaining the valid occlusion signals. The valid on / off status within each sampling period is then organized into a single set of data and stored continuously in chronological order of sampling time, forming an occlusion vector sequence that changes continuously with the progress of sampling time.

[0025] The extraction of the waveform signal showing the change in the number of continuously obstructed beams over time from the generated occlusion vector sequence is one of the core improvements of this solution compared to existing technologies. Existing conventional detection schemes only extract static occlusion data from a single frame, completely losing the dynamic characteristics of the occlusion event. In the preferred implementation, this step processes each set of data in the occlusion vector sequence frame by frame according to the sampling time sequence, extracting the number of continuously obstructed beams corresponding to that sampling time from each set of data. Then, the number of continuously obstructed beams corresponding to all sampling times is arranged chronologically to form a one-dimensional waveform signal that changes continuously over time. This fully preserves the dynamic temporal characteristics of the entire process of the occlusion event from its occurrence, development, to its end, providing core data support for accurately distinguishing between human intrusion risks and on-site interference.

[0026] The extracted real-time waveform signal is dynamically time-warped and compared with at least one pre-stored standard waveform template for human intrusion to calculate the waveform similarity between the real-time waveform and the standard template. Existing conventional time-series comparison schemes use a fixed time axis matching method, which cannot adapt to the waveform time length differences caused by different intrusion speeds and is prone to matching failure. In this optimized implementation, standard waveform templates corresponding to the entire process of a normal human intrusion into the protected area are pre-collected and stored. After obtaining the real-time waveform signal, the time axis correspondence between the real-time waveform and the standard template is adaptively adjusted through dynamic time warping comparison without changing the overall trend of the waveform. This effectively eliminates the time axis differences caused by slow, normal, and fast intrusions, finds the optimal matching relationship between the real-time waveform and the standard template, and then calculates the waveform similarity based on the degree of matching between the two. This similarity can accurately reflect the degree of matching between the current obstruction event and the human intrusion behavior.

[0027] The calculated waveform similarity is compared with a preset risk assessment threshold, and the corresponding anti-lifting detection result is output based on the comparison result. During the on-site optimization implementation process, the preset risk assessment threshold can be flexibly set according to the on-site protection level requirements and the degree of interference in the working environment. When the waveform similarity reaches or exceeds the preset threshold, the current obstruction event is determined to be a personnel intrusion event with anti-lifting risk, and the corresponding detection result with anti-lifting risk is output, which can be directly used to trigger subsequent safety protection actions. When the waveform similarity is lower than the preset threshold, the current obstruction event is determined to be an on-site interference event that will not cause the risk of lifting personnel, and the corresponding detection result with no anti-lifting risk is output, and the real-time acquisition and detection status is continuously returned, completing the closed loop of the entire detection process.

[0028] As a preferred embodiment of the above, the vertical detection range of the safety light curtain extends from the ground to the upper limit of the dangerous height at which a person can be suspended.

[0029] Specifically, in this preferred embodiment, the upper limit of the dangerous height for human lifting is set by taking into account the on-site working conditions of the copper wire drawing machine's wire laying frame, the average working height of adult personnel, the initial critical height of injury after being entangled and hooked by the copper rod, and the critical safe height for personnel to prevent falls and lifting in the general safety protection specifications for the metal wire processing industry. It is preferably selected as a height position of 1000mm above the working ground.

[0030] As a preferred embodiment of the above, the safety light curtain is a through-beam light curtain, with the transmitter and receiver of the safety light curtain respectively installed on both sides of the wire feeding frame, and the detection plane of the safety light curtain perpendicular to the ground.

[0031] Specifically, in the preferred embodiment, a through-beam light curtain is preferentially selected as the core detection device. Compared with reflective and diffuse reflection light curtains, the through-beam light curtain has separate transmitter and receiver installations, a stable beam transmission path, and stronger resistance to interference from dust, oil, and reflections from copper rods. It is also more suitable for the complex operating environment of copper wire drawing machines and wire feeding frames. The transmitter and receiver are preferably fixedly installed facing each other on both sides of the wire feeding frame along the horizontal width of the personnel working side. The installation height and horizontal installation position of the transmitter and receiver are completely corresponding, with no horizontal or vertical misalignment. This ensures that each beam emitted by the light curtain can be accurately captured by the corresponding receiver, avoiding detection blind spots or signal misjudgment due to installation misalignment. The detection plane of the light curtain is strictly perpendicular to the working ground and parallel to the direction of personnel entering the risk area. When personnel enter the risk area of ​​the cable tray from the working side, they will inevitably pass through the detection plane completely, forming a continuous beam blockage, which completely avoids the detection blind spot caused by personnel entering from the side or from a low position. At the same time, the perpendicular layout of the light curtain ensures that all beams of the light curtain are in the same vertical plane and will not intersect with the normal running path of the copper rod released by the cable tray. This avoids continuous blockage caused by the normal operation of the copper rod and provides a stable and accurate basic signal for subsequent acquisition of beam on / off status and generation of blockage vector sequence, further improving the stability and detection accuracy of the entire detection method.

[0032] As a preferred embodiment of the above, such as Figure 2 As shown, the on / off status of each beam is acquired in real time through a safety light curtain. The original signal is preprocessed to remove interference, generating a time-varying occlusion vector sequence, including: A10: Scan the received light intensity of each beam at a fixed sampling period, and determine the light intensity below the preset background threshold as an obstruction state; A20: Perform median filtering on the occlusion states of multiple consecutive frames, and arrange the occlusion states of each frame after filtering into a binary vector according to the beam order, and store it in chronological order as the occlusion vector sequence.

[0033] Specifically, in this preferred embodiment, a fixed sampling period of 1ms is preferentially selected to cyclically scan the received light intensity data corresponding to each beam of light in the safety light curtain. This sampling period can fully cover the occlusion change process in all scenarios, such as rapid entry and slow movement of personnel. It will not lose key temporal features due to excessively long sampling intervals, nor will it increase the computational load of the equipment due to redundant data generated by excessively high sampling frequencies. The preset background threshold is calibrated during the system initialization phase. The calibration process involves collecting the average received light intensity of each beam in an unobstructed state for 100 consecutive frames, and setting 30% of this average value as the background threshold for the corresponding beam. During the scanning process, when the real-time received light intensity of a beam is lower than the corresponding preset background threshold, the beam is determined to be in an obstructed state; otherwise, it is determined to be in an unobstructed state. This threshold setting method can adaptively offset the reference light intensity shift caused by changes in ambient light and slight dust accumulation on the light curtain lenses, significantly reducing the probability of misjudgment of the basic state.

[0034] For occlusion status data acquired from multiple consecutive frames, a median filter with a 3-frame sliding window is preferentially used. For each beam, the occlusion status corresponding to multiple consecutive frames is selected as the most frequently occurring status within the window as the effective status after filtering. This completely eliminates transient jump interference from a single beam in a single frame, including invalid interference signals such as reflections from the high-speed movement of the copper rod, drifting dust particles, and instantaneous beam offset caused by equipment vibration. At the same time, it does not change the continuous temporal characteristics of the actual occlusion event, avoiding the filtering process from erasing key action information of personnel intrusion. After filtering, the effective occlusion status of all beams in each frame is arranged in the vertical arrangement order of the beams from bottom to top, generating a set of binary vectors, where the occlusion status corresponds to the value 1 and the unoccluded status corresponds to the value 0. Then, the binary vectors corresponding to all sampling frames are stored sequentially in the order of sampling time, finally generating an occlusion vector sequence that changes over time.

[0035] For a specific example of field application, a safety light curtain is vertically arranged with 16 detection beams, numbered 1 to 16 in order from the ground to the upper limit of the danger height. In a certain sampling frame, beams numbered 3 to 5 are determined to be in an obstructed state, while the remaining beams are in an unobstructed state. This frame generates a corresponding 16-bit binary vector. By continuously sampling for 1000ms at a 1ms sampling period, 1000 sets of binary vectors arranged in chronological order can be obtained, which together form a complete obstruction vector sequence. This sequence can completely reconstruct the spatial location distribution and temporal changes of the obstruction event, providing accurate and stable basic data for subsequent waveform feature extraction.

[0036] As a preferred embodiment of the above, such as Figure 3 As shown, the waveform signal of the number of continuously obstructed beams changing over time is extracted from the obstruction vector sequence, including: B10: For each frame of occlusion vector, iterate through the occlusion state of each beam and calculate the number of the longest consecutively occluded beams as the occlusion quantity feature value of that frame. B20: Arrange the occlusion quantity feature values ​​of each frame into a one-dimensional sequence according to time order to obtain a waveform signal that changes with time.

[0037] Specifically, in this preferred embodiment, for each frame of occlusion vectors within the occlusion vector sequence, the occlusion status of all beams within a single frame is traversed sequentially from the lowest beam to the highest beam, following the vertical arrangement of beams from the ground to the upper limit of the danger height. Simultaneously, all continuously occluded beam segments within a single frame are counted, and the segment with the largest number of beams among all continuously occluded beam segments is selected. The number of beams corresponding to this segment is determined as the occlusion quantity feature value for that sampling frame. This calculation method differs from the conventional approach of directly counting the total number of occluded beams in a single frame. Its core advantage lies in the fact that when personnel enter the protected area, the light curtain formed by their bodies creates continuous and concentrated beam segments. However, the occlusions created by splashing materials, falling tools, swinging copper rods, and drifting dust are mostly scattered and discontinuous beams, or continuous beams with a length much shorter than the occlusion segments formed by the human body. By extracting the number of the longest continuous occlusion beams, scattered and invalid occlusion information within a single frame can be directly filtered out, preemptively eliminating most interference. Simultaneously, the core occlusion features formed by human intrusion are accurately preserved, significantly reducing the redundancy of subsequent data processing and improving feature recognition.

[0038] After calculating the occlusion quantity feature value for a single frame, the occlusion quantity feature values ​​corresponding to all sampled frames are arranged sequentially according to the sampling time, forming a one-dimensional temporal sequence. This sequence uses the sampling time as the horizontal axis and the number of longest consecutive occlusion beams corresponding to each frame as the vertical axis, ultimately generating a waveform signal that changes continuously over time. This waveform signal can completely reproduce the dynamic change trend of the entire process of the occlusion event from occurrence to development to end. For example, as a person enters the light curtain area and gradually enters the detection plane, the number of longest consecutive occlusion beams will gradually increase from 0, remain stable after the body is fully inside, and then gradually drop back to 0 when the person leaves. This completely preserves the temporal change characteristics of the person's intrusion action, providing a highly recognizable core data foundation for subsequent template matching and risk assessment.

[0039] As a preferred embodiment of the above, such as Figure 4 As shown, step S4 involves dynamically time-warping the waveform signal against at least one pre-stored standard waveform template for human intrusion, and calculating the waveform similarity, including: S41: Determine the time axis correspondence between the real-time waveform signal and the standard waveform template, and set the search window for dynamic time warping; S42: Calculate the distance between each point in the two sequences point by point, and construct the distance matrix using the cumulative distance recursive formula; S43: Search for the optimal alignment path from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; S44: Convert the minimum cumulative distance into a similarity score with a fixed range of values ​​through normalization.

[0040] Specifically, the process begins by aligning the start and end time axes of the real-time waveform and the standard waveform template, and simultaneously setting a search window with a width of 15% of the real-time waveform sequence length to balance matching accuracy and computational efficiency. Next, a matrix framework is constructed with each data point in the real-time waveform sequence as a row and each data point in the standard template as a column. The Euclidean distance between corresponding data points is calculated point by point as the base distance value. Starting from the matrix's origin, a cumulative distance recursive approach is used, taking the minimum cumulative distance between adjacent reachable positions and adding it to the current base distance value. This calculation is performed row by row and column by column to construct a complete distance matrix. Then, the process backtracks from the end point of the distance matrix to the origin, moving at each step towards the position with the minimum cumulative distance among adjacent reachable positions until returning to the origin, obtaining the optimal alignment path and the corresponding minimum cumulative distance between the two sequences. Finally, the minimum cumulative distance is normalized and converted into a fixed similarity score in the 0-1 range. Using the optimal alignment path length as the normalization benchmark, the final score is obtained by subtracting the ratio of the minimum cumulative distance to the path length from 1. The closer the score is to 1, the higher the matching degree with the standard template. For example, a normal-speed intrusion template with 200 pre-stored data points is used to collect real-time waveforms of 250 data points on-site. The minimum cumulative distance is calculated to be 120, the optimal path length is 260, and the final similarity score is 0.54, which can be directly used for subsequent threshold comparison.

[0041] As a preferred embodiment of the above, the human intrusion standard waveform template includes at least three templates corresponding to different intrusion speeds, which are used to match the intrusion waveforms of personnel entering at slow speed, normal speed, and fast speed into the light curtain area, respectively.

[0042] Specifically, in this preferred embodiment, the standard waveform templates for the three different speeds are all obtained by on-site measurement and standardization. During the collection process, adult workers of different heights and body types are invited to simulate the operation of the copper wire drawing machine and the laying frame. Various real personnel are invited to enter the scene. After collecting the corresponding waveforms, they are deduplicated, smoothed and averaged to form standard templates. These templates are completely consistent with the actual on-site working conditions, rather than theoretical simulation data, thus ensuring the accuracy of template matching.

[0043] The slow entry template corresponds to a personnel walking speed of approximately 0.3-0.8 m / s, suitable for low-speed intrusion scenarios where personnel are entangled and dragged by copper poles or slowly step into the risk area while bending over; the normal speed entry template corresponds to a personnel walking speed of approximately 0.8-1.5 m / s, suitable for normal scenarios where personnel walk and work normally into the risk area, which is also the most common personnel intrusion situation on site; the fast entry template corresponds to a personnel walking speed of approximately 1.5-3 m / s, suitable for extremely high-risk scenarios where personnel rush in urgently and are quickly hooked and pulled into the risk area by high-speed copper poles.

[0044] In the actual detection process, the extracted real-time waveform signal is compared with the three types of templates through dynamic time warping. The highest waveform similarity score among the three comparison results is taken as the final judgment basis, ensuring that no matter what speed the personnel enter the risk area, the corresponding optimal matching template can be found, avoiding matching failure, missed detection and misjudgment caused by a single template being unable to adapt to different speeds.

[0045] As a preferred embodiment of the above, the waveform similarity is compared with a preset threshold, and the corresponding anti-hanging detection result is output, including: The calculated waveform similarity is compared with a preset similarity threshold. If the waveform similarity reaches or exceeds the similarity threshold, the current occlusion event is determined to match the human intrusion standard waveform template, and a detection result indicating a risk of occlusion is output. If the waveform similarity is lower than the similarity threshold, the output will show a detection result indicating no risk of being lifted.

[0046] Specifically, this preferred embodiment achieves accurate binary judgment of personnel intrusion risk through a clear threshold comparison logic, providing a clear and reliable triggering basis for the safety protection actions of copper wire drawing machines, and is fully compatible with the execution logic of industrial site safety control systems. The preset similarity threshold is determined by combining the protection level requirements of on-site operations, the degree of environmental interference, and the safety protection standards of the metal wire processing industry, and is calibrated through a large amount of on-site measurement data. It is initially set to 0.75, while also supporting flexible adjustment according to the actual on-site working conditions. In high-risk operation scenarios, the threshold can be appropriately lowered to avoid missed detections, while in complex scenarios with many interference sources, the threshold can be appropriately increased to reduce false triggers, demonstrating strong on-site adaptability. During the actual comparison process, the highest waveform similarity obtained after comparing the real-time waveform signal with all human intrusion standard waveform templates is taken as the effective comparison value and compared with the preset threshold to ensure coverage of all possible personnel intrusion speed scenarios. When the effective waveform similarity reaches or exceeds the preset similarity threshold, the current obstruction event is determined to be highly matched with the action characteristics of human intrusion, belonging to a personnel intrusion event with anti-lifting risk. The corresponding output is a detection result indicating anti-lifting risk. This result can directly trigger the emergency stop of the copper wire drawing machine's pay-off frame, on-site audible and visual alarms, and other safety protection actions to block the risk at the first time. When the effective waveform similarity is lower than the preset similarity threshold, the current obstruction event is determined to be an on-site interference event that will not cause the risk of lifting personnel, such as material splashing, tool falling, or copper rod swinging. The corresponding output is a detection result indicating no anti-lifting risk. The system continuously returns the real-time acquisition and detection status, forming a complete detection closed loop.

[0047] Example 2: Based on the same inventive concept as the copper wire drawing machine anti-slip detection method based on light curtain recognition in the foregoing embodiments, the present invention also provides a copper wire drawing machine anti-slip detection system based on light curtain recognition, comprising: The detection baseline establishment module performs area calibration and initialization of the safety light curtain for the risk area of ​​the copper wire drawing machine's pay-off frame, and establishes the detection benchmark. The vector sequence generation module collects the on / off status of each beam in real time through the safety light curtain, preprocesses the original signal to remove interference, and generates an occlusion vector sequence that changes over time. The waveform signal extraction module extracts the waveform signal of the number of continuously blocked beams changing over time from the blockage vector sequence; The similarity calculation module performs dynamic time-normalization comparison between the waveform signal and at least one pre-stored standard waveform template for human intrusion, and calculates the waveform similarity. The detection result generation module compares the waveform similarity with a preset threshold and outputs the corresponding anti-hanging detection result.

[0048] The detection system described above in this invention can effectively realize the anti-slip detection method for large copper wire pulling machines based on light curtain recognition. The technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0049] As a preferred embodiment of the above, the similarity calculation module includes: The search window setting unit determines the time axis correspondence between the real-time waveform signal and the standard waveform template, and sets the search window for dynamic time warping. The distance matrix construction unit calculates the distance between each point in the two sequences point by point, and constructs the distance matrix using the cumulative distance recursive formula; The optimal path search unit searches for the optimal alignment path from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; The similarity normalization unit converts the minimum cumulative distance into a similarity score with a fixed value range through normalization processing.

[0050] Similarly, the above-mentioned optimization schemes for the system can also achieve the optimization effects corresponding to the methods in Embodiment 1, which will not be repeated here.

[0051] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering 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. Thus, 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. A method for detecting anti-slip issues in large copper wire drawing machines based on light curtain recognition, characterized in that, The method includes: For the risk area of ​​the copper wire drawing machine's pay-off frame, the area of ​​the safety light curtain is calibrated and initialized to establish a detection benchmark; The on / off status of each beam is collected in real time by a safety light curtain, the original signal is preprocessed to remove interference, and a time-varying occlusion vector sequence is generated. Extract the waveform signal of the number of continuously blocked beams changing over time from the blockage vector sequence; The waveform signal is dynamically time-normalized and compared with at least one pre-stored standard waveform template for human intrusion to calculate the waveform similarity. The waveform similarity is compared with a preset threshold, and the corresponding anti-hanging detection result is output.

2. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The vertical detection range of the safety light curtain extends from the ground to the upper limit of the dangerous height at which a person can be suspended.

3. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The safety light curtain is a through-beam light curtain, with the transmitter and receiver of the safety light curtain respectively installed on both sides of the wire feeding frame, and the detection plane of the safety light curtain is perpendicular to the ground.

4. The method for detecting anti-slipping of large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The on / off status of each beam is acquired in real time through a safety light curtain. The original signal is preprocessed to remove interference, generating a time-varying occlusion vector sequence, including: The received light intensity of each beam is scanned at a fixed sampling period, and those below a preset background threshold are judged as occlusion states. Median filtering is applied to the occlusion states of multiple consecutive frames, and the occlusion states of each frame after filtering are arranged into binary vectors according to the beam order, and stored in chronological order as the occlusion vector sequence.

5. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, Extracting the waveform signal of the number of continuously blocked beams changing over time from the blocking vector sequence includes: For each frame of occlusion vector, traverse the occlusion state of each beam and calculate the number of the longest consecutively occluded beams as the occlusion quantity feature value of that frame. Arrange the occlusion quantity feature values ​​of each frame into a one-dimensional sequence according to time order to obtain the waveform signal that changes with time.

6. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The waveform signal is dynamically time-normalized and compared with at least one pre-stored standard waveform template for human intrusion to calculate waveform similarity, including: Determine the time axis correspondence between the real-time waveform signal and the standard waveform template, and set the search window for dynamic time warping; Calculate the distance between each point in the two sequences point by point, and construct the distance matrix using the cumulative distance recursive formula; The optimal alignment path is searched from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; The minimum cumulative distance is converted into a similarity score with a fixed value range through normalization.

7. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The human intrusion standard waveform template includes at least three templates corresponding to different intrusion speeds, which are used to match the intrusion waveforms of personnel entering the light curtain area at slow speed, normal speed, and fast speed.

8. The method for detecting anti-slipping issues in large copper wire drawing machines based on light curtain recognition according to claim 1, characterized in that, The waveform similarity is compared with a preset threshold, and the corresponding anti-hanging detection results are output, including: The calculated waveform similarity is compared with a preset similarity threshold. If the waveform similarity reaches or exceeds the similarity threshold, it is determined that the current occlusion event matches the human intrusion standard waveform template, and a detection result indicating a risk of shoving is output. If the waveform similarity is lower than the similarity threshold, a detection result indicating no risk of suspension is output.

9. A copper wire drawing machine anti-slip detection system based on light curtain recognition, characterized in that, The system includes: The detection baseline establishment module performs area calibration and initialization of the safety light curtain for the risk area of ​​the copper wire drawing machine's pay-off frame, and establishes the detection benchmark. The vector sequence generation module collects the on / off status of each beam in real time through the safety light curtain, preprocesses the original signal to remove interference, and generates an occlusion vector sequence that changes over time. The waveform signal extraction module extracts the waveform signal of the number of continuously blocked beams changing over time from the blockage vector sequence; The similarity calculation module performs dynamic time-normalization comparison between the waveform signal and at least one pre-stored human intrusion standard waveform template to calculate the waveform similarity. The detection result generation module compares the waveform similarity with a preset threshold and outputs the corresponding anti-hanging detection result.

10. The copper wire drawing machine anti-slip detection system based on light curtain recognition according to claim 9, characterized in that, The similarity calculation module includes: The search window setting unit determines the time axis correspondence between the real-time waveform signal and the standard waveform template, and sets the search window for dynamic time warping. The distance matrix construction unit calculates the distance between each point in the two sequences point by point, and constructs the distance matrix using the cumulative distance recursive formula; The optimal path search unit searches for the optimal alignment path from the start point to the end point of the distance matrix to obtain the minimum cumulative distance; The similarity normalization unit converts the minimum cumulative distance into a similarity score with a fixed value range through normalization processing.

Images