A method and system for detecting the compression performance of an elevator polyurethane buffer

Abstract

The present application relates to the elevator technical field, especially a kind of elevator polyurethane bumper compression performance detection method and system.The method includes the following steps: camera obtains the video of crash plate and polyurethane bumper, YOLOv8 network model and DeepSORT algorithm carry out interframe target tracking to obtain the pixel-level displacement of crash plate, displacement sensor measures the physical displacement of crash plate in real time, constructs the continuous displacement curve of polyurethane bumper compression rebound.Based on the fusion index of energy loss rate, dynamic damping ratio and the coupling contrast function of compression rebound, whether polyurethane bumper exists compression rebound hysteresis or compression rebound asymmetric anomaly is judged;The detection report of elevator polyurethane bumper is generated and uploaded to cloud platform or server.The method accurately judges whether polyurethane bumper exists compression rebound hysteresis or compression rebound asymmetric anomaly, system equipment integration is high, portable and can adapt to the space limitation and real-time requirement of elevator field test environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elevators, and in particular to a method and system for detecting the compression performance of an elevator polyurethane buffer. BACKGROUND

[0002] With the rapid development of the elevator industry, intelligent detection of elevator safety performance is increasingly valued, especially in the automated detection of elevator buffers. Currently, some domestic and foreign research institutions and enterprises have initially explored related technologies, mainly using image recognition technology to fuse sensor data to identify and record the state of the buffer, and to realize the automation of the test process through single-chip control. For example, CN112712517A discloses a hydraulic buffer detection system and method based on image intelligent recognition, in which a camera automatically captures images of the hydraulic buffer test process at a set cycle time, extracts the edge features in the images and calculates the compression amount, and obtains the corresponding data of time and movement distance in the entire detection process. CN118762244A discloses a hydraulic buffer detection system and method based on image recognition, in which a convolutional neural network (CNN) is used to preprocess the image of the elevator hydraulic buffer, extract the appearance defects, wear degree and deformation, and calculate the weight of the appearance data and displacement data to evaluate the performance, and also calculate the wear rate of the buffer to predict the service life and maintenance requirements of the hydraulic buffer.

[0003] The above technology is for the automated detection of hydraulic buffers, and polyurethane buffers are increasingly widely used. The current elevator supervision and periodic inspection rules require that elevators with non-metallic material and non-linear energy storage type buffers perform compression tests, i.e., polyurethane buffer detection of elevators requires compression tests. The above-mentioned buffer detection method is not suitable for detecting the compression rebound hysteresis or compression rebound asymmetry abnormalities of polyurethane buffers.

[0004] There are automatic detections of elevator polyurethane buffers in the prior art, such as the elevator polyurethane buffer detection method based on machine vision disclosed in CN117735357A. The un-compressed state and the compressed state of the polyurethane buffer are extracted by the CCD vision sensing device, and the shape of the polyurethane buffer is compared with the preset standard polyurethane buffer shape state for judgment. According to the running speed of the elevator, the time when the polyurethane buffer is contacted is determined, and a signal is sent to the CCD vision sensing device to make it take a picture. This scheme only detects the shape of the polyurethane buffer, and it is difficult to obtain the key performance of the compression rebound hysteresis and the compression rebound asymmetry of the polyurethane buffer based on the shape of the polyurethane buffer. In the prior art, the compression video of the polyurethane buffer is taken, and whether the rebound hysteresis and the rebound asymmetry exist is observed by manually observing the video playback. There is a subjective error. Therefore, it is necessary to provide a method and system for accurately detecting the compression performance of the elevator polyurethane buffer. SUMMARY

[0005] The purpose of the present application is to provide a method and system for detecting the compression performance of an elevator polyurethane buffer, which can accurately determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry, has high system device integration, is portable, and can adapt to the space limitations and real-time requirements of the elevator field test environment.

[0006] To achieve this purpose, the following technical solutions are used in the present application:

[0007] A method for detecting the compression performance of an elevator polyurethane buffer, which is applied to an elevator polyurethane buffer compression performance detection system. The system includes a camera, a displacement sensor, an edge module, a touch screen, and a communication module. The edge module is equipped with a YOLOv8 network model and a DeepSORT algorithm, and the touch screen is used for inputting device information.

[0008] The method includes the following steps:

[0009] S1: The camera acquires the video of the striker plate and the polyurethane buffer during the compression process and the rebound process of the polyurethane buffer, and the edge module pre-processes the image frames.

[0010] S2: The pre-processed image frames are input into the trained YOLOv8 network model for target recognition. Based on the target recognition result, the DeepSORT algorithm is used for inter-frame target tracking to obtain the pixel-level displacement of the striker plate. The physical displacement of the striker plate is measured in real time by the displacement sensor.

[0011] S3: The pixel-level displacement and the physical displacement are fused to construct a continuous displacement curve of the polyurethane buffer compression rebound.

[0012] S4: based on the continuous displacement curve of the polyurethane buffer compression rebound, calculating the energy loss rate of the striker plate and the dynamic damping ratio of the polyurethane buffer, and constructing the coupling contrast function of the compression rebound; based on the energy loss rate, the dynamic damping ratio and the coupling contrast function of the compression rebound, judging whether the polyurethane buffer exists compression rebound hysteresis or compression rebound asymmetry abnormality;

[0013] S5: based on the continuous displacement curve, the abnormality judgment result and the equipment information, generating the detection report of the polyurethane buffer of the elevator;

[0014] S6: the communication module uploads the detection report to the cloud platform or the server.

[0015] Further, in the step S2, the measured physical displacement is dynamically compensated based on the material characteristics correction of the environmental factors, comprising:

[0016] S201: establishing a nonlinear temperature compensation model:

[0017] Wherein, D a is the physical displacement after temperature compensation, D raw is the measured physical displacement, α1, α2, α3 is the first-order, second-order and third-order temperature drift coefficient, is the temperature difference coefficient, K material () is the elastic modulus correction factor of the polyurethane material with temperature change, ζ1, ζ2, ζ3 is the aging attenuation coefficient, N is the cumulative compression times, N0 is the rated life times, is the cumulative times coefficient, , is the use time coefficient, , t is the used time, t0 is the design life time, is the load action coefficient, , is the stress corresponding to the rated load, is the working stress at the moment, is the cumulative stress time integral;

[0018] S202: establishing a humidity-temperature coupling compensation model: Wherein, D b is the physical displacement after temperature and humidity coupling compensation, β1 and β2 are the first-order and second-order humidity influence coefficients, ρ is the temperature and humidity interaction coefficient, H is the humidity coefficient, , is the reference calibration humidity value, is the relative humidity;

[0019] S203: calculating the physical displacement after compensation: ​.

[0020] Furthermore, step S3 includes,

[0021] S301: Convert pixel-level displacement to physical displacement: , where L vision (t) represents the calculated physical displacement. This represents the target motion quantity in the pixel domain. This represents the mapping coefficient from pixel to physical displacement obtained from system calibration;

[0022] S302: Interpolation is used to align physical displacement with pixel-level displacement in time. , where L tof (t s ) represents the time t to be aligned. s The physical displacement, t a and t b Distance t s The two most recent sampling times, L tof (t a ) is t a Physical displacement at the sampling time, L tof (t b ) is t b Physical displacement at the sampling time;

[0023] The calculated physical displacement and the time-aligned physical displacement are merged to generate a continuous displacement curve for the compression rebound of the polyurethane buffer: Where λ1 and λ2 are adaptive weighting coefficients, satisfying λ1+λ2=1;

[0024] S303: The fused data is processed using the Kalman filter algorithm to reconstruct the smooth trajectory of the continuous displacement curve.

[0025] Furthermore, step S4 includes:

[0026] S410: Construct an energy estimation model, dividing the entire compression and rebound cycle of the polyurethane buffer into five stages: contact period, linear compression period, nonlinear compression period, rebound initiation period, and full rebound period, and calculate the energy loss rate of the impact plate in each stage. ;

[0027] S420: Based on the stages defined in step S410, calculate the dynamic damping ratio for each stage. ;

[0028] S430: Based on the stages divided in step S410, calculate the compression rate curves for the linear compression period and the nonlinear compression period, and calculate the rebound rate curves for the rebound initiation period and the full rebound period, respectively. Construct a coupling comparison function S for the compression rate curve and the rebound rate curve. CR ; Calculate the coupling comparison function S of the compression velocity curve during the linear compression period and the rebound velocity curve during the full rebound period. CR1 And the coupled comparison function S for calculating the compression velocity curve during the nonlinear compression period and the rebound velocity curve during the rebound initiation period. CR2 ;

[0029] S440: Constructing the evaluation model: ;

[0030] Where E is the overall score, and w1, w2, and w3 are the weights. The average energy loss rate represents the normal sample. The standard deviation of the energy loss rate represents the normal sample. The mean dynamic damping ratio represents the normal sample. The standard deviation of the dynamic damping ratio represents the normal sample. The mean coupling score representing the normal sample. The standard deviation of the coupling score represents the normal sample; when calculating the comprehensive score E during the contact period, let... When calculating the combined score E for the linear compression period and the full rebound period, let S... CR =S CR1 When calculating the combined score E for the nonlinear compression period and the rebound initiation period, let S... CR =S CR2 ;

[0031] The overall score E for each stage is compared with the corresponding threshold. If the overall score E for any stage exceeds the corresponding threshold... When this occurs, it is determined that there is a compression rebound lag or a compression rebound asymmetry anomaly.

[0032] Furthermore, step S410 includes:

[0033] S411: Calculate the kinetic energy of the impact plate. and potential energy :

[0034] m is the mass of the impact plate. It is the derivative of the displacement s(t) of the impact plate with respect to time t;

[0035] g is the acceleration due to gravity, and h(t) is the velocity of the impact plate relative to s. ref The height difference between the positions, s(t) is the instantaneous displacement of the impact plate at any time t, sref This is the position of the polyurethane buffer when it is uncompressed and at rest.

[0036] S412: Calculate the energy dissipation rate , This indicates that the system energy is being dissipated. Based on the energy dissipation rate, the entire compression-rebound cycle is divided into five stages: the contact period, the linear compression period, the nonlinear compression period, the rebound initiation period, and the complete rebound period.

[0037] S413: Switching thresholds for five stages , It is a set of segmented switching time points for threshold determination, wherein the time point when the threshold is first reached or exceeded within a stage is taken as the segmented switching time point;

[0038] Energy loss rate Λ at each stage k The calculation is as follows: k=1, hour, It is the starting moment. This refers to the moment when video recording begins in step S1; k=5. hour, It's the finish line, the finish line. This refers to the moment when video recording ends in step S1.

[0039] Furthermore, step S420 includes:

[0040] S421: Obtain the displacement peak values ​​for each stage from the continuous displacement curve. These displacement peak values ​​constitute a sequence of displacement oscillation peak values ​​within each stage. Calculate the displacement oscillation decrease in each stage. : , It is the number of displacement peaks;

[0041] S422: Calculate the damping ratio using the time-domain method at each stage: , p The logarithmic decay rate is used; then the damping ratio is calculated using frequency domain analysis. The dominant frequency fr and half-power bandwidth Δfr in the entire continuous displacement curve are extracted by Fourier transform.

[0042] The stage dynamic damping ratio is obtained by weighted fusion. , ，c t , c f It's weight. c t , c f ∈[0,1].

[0043] Furthermore, step S430 includes:

[0044] S431: Calculate the compression rate curve during the linear compression period. , S c2 It is the displacement during the linear compression period obtained based on the continuous displacement curve;

[0045] Calculate the compression rate curve during the nonlinear compression period. , S c3 It is the displacement during the nonlinear compression period obtained based on the continuous displacement curve;

[0046] Calculate the rebound speed curve during the rebound initiation phase. , S r4 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve;

[0047] Calculate the rebound speed curve during the full rebound period. , S r5 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve;

[0048] S432: Based on the cosine similarity principle, design a coupling comparison function S for the compression velocity curve and the rebound velocity curve. CR : S CR The value range is [0, 1], S CR The closer the value is to 1, the more consistent the shape and phase of the compression velocity curve and the rebound velocity curve are. for or , for or ;

[0049] Based on the coupling comparison function S CR ,calculate and Coupling comparison function S CR1 and calculation and Coupling comparison function S CR2 .

[0050] Furthermore, the method for handling the video of the impact plate and polyurethane buffer in step S1 includes:

[0051] When the displacement sensor measures that the physical displacement of the impact plate is less than the set threshold for the first time and the duration exceeds the threshold frame length, it is determined that the impact plate is about to make physical contact with the polyurethane buffer. This moment is marked as the starting point of the compression process, and video recording is triggered.

[0052] When the displacement sensor measures that the physical displacement of the impact plate is greater than the set threshold and the duration exceeds the set time, the video recording ends and the video is saved.

[0053] Step S2 obtains the pixel-level displacement of the impact plate in the recorded video.

[0054] Furthermore, in step S1, the method for preprocessing the image frame includes:

[0055] S101: Lens distortion correction: Based on the camera intrinsic parameter matrix and distortion coefficients, geometric correction is performed using OpenCV's undistort() function to eliminate radial and tangential distortion caused by wide-angle lenses;

[0056] S102: Image contrast enhancement: Convert the image to the LAB color space, perform adaptive histogram equalization on the brightness channel, and enhance the local contrast and edge features of the image.

[0057] S103: Region of Interest Extraction: Based on the prior position of the buffer in the image or the result of dynamic target localization, crop the central region of the image or a specified region as the region of interest.

[0058] S104: Image size normalization: Scales the cropped image frame to the set pixel size.

[0059] Accordingly, the present invention also provides an elevator polyurethane buffer compression performance testing system, which is used to implement the above-mentioned elevator polyurethane buffer compression performance testing method.

[0060] The system includes a housing and a camera, displacement sensor, edge module, touch screen and communication module mounted on the housing. The YOLOv8 network model and the DeepSORT algorithm are deployed on the edge module.

[0061] The camera is used to record video of the impact plate and the polyurethane buffer during the compression and rebound processes of the polyurethane buffer.

[0062] The displacement sensor is used to measure the physical displacement of the impact plate in real time;

[0063] The edge module is used to preprocess image frames, obtain pixel-level displacement of the impact plate, and determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly and generate a detection report.

[0064] The touchscreen is used to play videos and to input device information;

[0065] The communication module is used to upload the test report to the cloud platform or server.

[0066] The technical solution provided by this invention may include the following beneficial effects:

[0067] 1. The pixel-level displacement of the impact plate is obtained by combining the YOLOv8 network model and the DeepSORT algorithm, which avoids the lack of adaptability to typical elevator site environments such as changes in lighting, shadow interference, and complex backgrounds; the pixel-level displacement and physical displacement are fused to construct a continuous displacement curve of the compression and rebound of the polyurethane buffer, thereby improving the accuracy of the target displacement data.

[0068] 2. Based on accurate displacement data, the energy loss rate, dynamic damping ratio, and compression rebound coupling comparison function are calculated to determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly, so as to achieve accurate evaluation of the compression performance of the polyurethane buffer and solve the problem of medium error in judging compression performance by manual observation of video in the existing technology.

[0069] 3. Video recording and compression status detection form a complete closed loop: Simultaneously realize video recording of the compression process, target tracking and status comparison analysis to ensure data integrity and traceability, and meet the requirements of current elevator inspection regulations for the objectivity and traceability of test records.

[0070] 4. The system of the present invention has a high degree of integration. The camera, displacement sensor, edge module, touch screen and communication module can be integrated into a small housing, which is portable and can adapt to the space constraints and real-time requirements of the elevator on-site inspection environment. Attached Figure Description

[0071] Figure 1 This is a flowchart of one embodiment of the present invention;

[0072] Figure 2 This is a flowchart of another embodiment of the present invention;

[0073] Figure 3 This is a schematic diagram of the system of the present invention when the camera is in working condition;

[0074] Figure 4 This is a schematic diagram of the system of the present invention when the camera is in the retracted state. Detailed Implementation

[0075] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0076] When the elevator car reaches the bottom of the pit, the impact plate at the bottom of the car disengages from the polyurethane buffer. The car continues to descend, forcing the polyurethane buffer to compress. When the buffer reaches its lowest point, it rebounds. During the compression and rebound process, the polyurethane buffer remains in contact with the impact plate; therefore, the position of the impact plate is the top position of the polyurethane buffer. In this invention, the top position of the polyurethane buffer is determined by detecting the position of the impact plate. The following describes the process in conjunction with... Figures 1 to 2 This invention describes a method for testing the compression performance of an elevator polyurethane buffer according to an embodiment of the present invention. The method is applied to an elevator polyurethane buffer compression performance testing system, which includes a camera, a displacement sensor, an edge module, a touch screen, and a communication module. The edge module is equipped with a YOLOv8 network model and a DeepSORT algorithm, and the touch screen is used to input device information.

[0077] The method includes the following steps:

[0078] S1: The camera acquires video of the collision plate and the polyurethane buffer during the compression and rebound processes of the polyurethane buffer, and the edge module preprocesses the image frames.

[0079] S2: Input the preprocessed image frames into the trained YOLOv8 network model to perform target recognition on the images; based on the target recognition results, perform inter-frame target tracking using the DeepSORT algorithm to obtain the pixel-level displacement of the impact plate; measure the physical displacement of the impact plate in real time using a displacement sensor;

[0080] S3: The pixel-level displacement and physical displacement are fused to construct a continuous displacement curve of the compression rebound of the polyurethane buffer;

[0081] S4: Based on the continuous displacement curve of the compression rebound of the polyurethane buffer, calculate the energy loss rate of the impact plate and the dynamic damping ratio of the polyurethane buffer, and construct the coupling comparison function of compression rebound; based on the energy loss rate, dynamic damping ratio and the coupling comparison function of compression rebound, determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly.

[0082] S5: Generate an inspection report for the elevator polyurethane buffer based on the continuous displacement curve, anomaly judgment results, and equipment information;

[0083] S6: The communication module uploads the test report to the cloud platform or server.

[0084] This invention combines a YOLOv8 network model and the DeepSORT algorithm to obtain pixel-level displacement of the impact plate, avoiding insufficient adaptability to typical elevator environments such as lighting changes, shadow interference, and complex backgrounds. It integrates pixel-level and physical displacement to construct a continuous displacement curve for the compression rebound of the polyurethane buffer, improving the accuracy of the target displacement data. Based on accurate displacement data, this scheme calculates the energy loss rate, dynamic damping ratio, and compression rebound coupling comparison function to determine whether the polyurethane buffer exhibits compression rebound hysteresis or asymmetric anomalies, achieving accurate evaluation of the polyurethane buffer's compression performance and solving the error problem of manually observing video to judge compression performance in existing technologies. This polyurethane buffer detection method provides a reference for a unified data standard and evaluation system in the current field, facilitating its application and promotion.

[0085] This solution integrates three dimensions: energy loss rate characteristics, dynamic damping dynamic characteristics, and time-series comparison anomaly detection behavior characteristics of compression-rebound full-cycle coupling behavior. It breaks through the bottleneck of traditional single physical quantity or static threshold detection and realizes multi-level intelligent identification from local stage anomalies to full-cycle behavioral anomalies.

[0086] It should be noted that the YOLOv8 network model in this invention is trained using a dataset consisting of pre-collected polyurethane buffer detection images to improve the model's recognition accuracy.

[0087] The displacement sensor of this invention is a Time-of-Flight (ToF) laser ranging sensor. Because the response characteristics of a ToF laser ranging sensor exhibit nonlinear drift under different ambient temperatures, the output raw ranging value will generate systematic errors with temperature changes. Simultaneously, the physical properties of the polyurethane buffer material itself (such as its elastic modulus) also fluctuate with temperature. Therefore, in one embodiment of this invention, the measured data, after joint correction for temperature drift and material properties, can accurately reflect the actual compression stroke of the buffer. Specifically, in step S2, the measured physical displacement is dynamically compensated for environmental factors based on material property correction to correct systematic errors caused by temperature and humidity changes, avoiding hysteresis misjudgments caused by small displacement deviations, thereby ensuring high accuracy and stability of the detection results. This includes:

[0088] S201: Establish a nonlinear temperature compensation model:

[0089] ;

[0090] Among them, D a It is the physical displacement after temperature compensation, D raw These are the measured physical displacements, and α1, α2, and α3 are the first-, second-, and third-order temperature drift coefficients, respectively. It is the temperature difference coefficient, derived from Normalization yields, This represents the maximum permissible temperature difference (e.g., ±30°C), T is the real-time ambient temperature (collected in real-time by a temperature sensor), T0 is the reference calibration temperature, and K... material () represents the temperature-dependent modulus correction factor for polyurethane materials, ζ1, ζ2, and ζ3 are aging degradation coefficients, N is the cumulative compression cycles, and N0 is the rated life cycles. It is the cumulative frequency coefficient. , It uses a time coefficient. t is the elapsed time, and t0 is the design life. It is the load application factor. , The stress is the stress corresponding to the rated load, σ(τ) is the working stress at time τ, and ∫σ(τ)dτ is the cumulative stress time integral; the cumulative compression count N is the input data at the start of the test. This data comes from the historical database of special equipment testing, which records the number of compression counts during buffer testing and the number of compression counts during elevator malfunctions. The working stress σ(τ) at time τ is calculated as follows: First, the acceleration is obtained based on the physical displacement change of the displacement sensor. The force is calculated using the acceleration and the mass of the impact plate. Then, the working stress (the pressure per unit area on the top surface of the polyurethane buffer at time τ) is obtained using the contact area between the polyurethane buffer and the impact plate.

[0091] The nonlinear temperature compensation model in this scheme is formed by combining and normalizing existing sensor temperature drift compensation, modulus correction models in materials mechanics, and the exponential decay law theory of fatigue life.

[0092] S202: Establish a humidity-temperature coupled compensation model: , where D b This represents the physical displacement after temperature and humidity coupling compensation, β1 and β2 are the first and second order influence coefficients of humidity, ρ is the temperature and humidity interaction coefficient, and H is the humidity coefficient. , This is the reference calibration humidity value. It refers to relative humidity;

[0093] S203: Calculate the compensated physical displacement: .

[0094] In one embodiment of the present invention, pixel-level displacement and physical displacement are fused into a high-precision, time-consistent continuous displacement curve through algorithms such as time synchronization and interpolation registration. This method fully leverages the complementary advantages of the two types of sensors, improves the spatial and temporal resolution of compressed data, and provides high-precision data support for modeling the compression process. Specifically, step S3 includes:

[0095] S301: Convert pixel-level displacement into physical displacement , where L vision (t) represents the calculated physical displacement, Δu(t) represents the target motion in the pixel domain, and γ represents the mapping coefficient from pixel to physical displacement (e.g., millimeters) obtained by system calibration.

[0096] S302: Interpolation is used to align physical displacement with pixel-level displacement in time, that is, to register physical displacement data onto a time axis consistent with pixel displacement data to achieve time alignment. , where L tof (t s ) represents the time t to be aligned. s The physical displacement, t a and t b Distance t s The two most recent sampling times, L tof (t a ) is t a Physical displacement at the sampling time, L tof (t b ) is t b Physical displacement at the sampling time;

[0097] The calculated physical displacement and the time-aligned physical displacement are merged to generate a continuous displacement curve for the compression rebound of the polyurethane buffer: Where λ1 and λ2 are adaptive weighting coefficients, satisfying λ1+λ2=1;

[0098] S303: The fused data is processed using a Kalman filter algorithm to reconstruct a smooth trajectory of the continuous displacement curve. In this scheme, an extended Kalman filter (EKF) is applied to the sequence of fused displacement and velocity data to suppress data fluctuations caused by visual detection errors, ranging sensor noise, and scene disturbances, thereby achieving smooth trajectory reconstruction of the impact plate.

[0099] Furthermore, step S4 includes:

[0100] S410: Construct an energy estimation model, combining displacement data from the continuous displacement curve, and divide the entire compression and rebound cycle of the polyurethane buffer into five stages: contact period, linear compression period, nonlinear compression period, rebound initiation period, and full rebound period. Calculate the energy loss rate of the impact plate in each stage. ;

[0101] S420: Based on the stages defined in step S410, calculate the dynamic damping ratio for each stage. ;

[0102] S430: Based on the stages divided in step S410, calculate the compression rate curves for the linear compression period and the nonlinear compression period, and calculate the rebound rate curves for the rebound initiation period and the full rebound period, respectively. Construct a coupling comparison function S for the compression rate curve and the rebound rate curve. CR ; Calculate the coupling comparison function S of the compression velocity curve during the linear compression period and the rebound velocity curve during the full rebound period. CR1 And the coupled comparison function S for calculating the compression velocity curve during the nonlinear compression period and the rebound velocity curve during the rebound initiation period. CR2 ;

[0103] S440: Constructing the evaluation model: ;

[0104] Where E is the overall score, and w1, w2, and w3 are the weights. The average energy loss rate represents the normal sample. The standard deviation of the energy loss rate represents the normal sample. The mean dynamic damping ratio represents the normal sample. The standard deviation of the dynamic damping ratio represents the normal sample. The mean coupling score representing the normal sample. The standard deviation of the coupling score represents the normal sample; when calculating the comprehensive score E during the contact period, let... When calculating the combined score E for the linear compression period and the full rebound period, let S... CR =S CR1 When calculating the combined score E for the nonlinear compression period and the rebound initiation period, let S... CR =S CR2 ;

[0105] The overall score E for each stage is compared with the corresponding threshold. If the overall score E for any stage exceeds the corresponding threshold... At this point, it is determined that there is a compression rebound hysteresis or compression rebound asymmetry anomaly. Threshold Thresholds are obtained based on routine inspection experience.

[0106] This scheme, through steps S410-S430, achieves precise analysis of the compression and rebound process data of the polyurethane buffer, and employs a multi-dimensional joint evaluation method, namely, the fusion of multi-dimensional physical characteristics, to determine the existence of compression-rebound hysteresis or compression-rebound asymmetry anomalies. First, the energy loss rate reflects the energy loss changes of the polyurethane material at each stage, sensitively revealing the abnormal energy accumulation during rebound hysteresis and the energy loss differences during asymmetric rebound. Second, the dynamic damping ratio reflects the damping characteristics of the material's oscillations, revealing the changes in viscoelastic properties caused by hysteresis or asymmetry. Furthermore, the coupling comparison function S... CRThis indicator measures the correspondence between the compression and rebound processes in terms of energy progress. By comparing the matching degree of the two velocity curves in the energy release and absorption process, it can intuitively reveal the overall coordination and symmetry between compression and rebound. This indicator can comprehensively reflect the overall coordination and symmetry of the buffer in the entire cycle of energy flow, making up for the inability of a single local feature to reveal global anomalies. Energy dissipation, dynamic damping ratio, and compression-rebound coupling consistency scores are assigned corresponding "similarity scores," and finally, they are weighted and summarized into a total score E. If the score corresponding to a certain stage is too low, or the total score E does not meet the standard, it is determined that the buffer has an anomaly of "rebound hysteresis" or "return asymmetry." At the same time, it can also mark at which stage the problem occurred.

[0107] Furthermore, this invention estimates the kinetic energy change and potential energy conversion corresponding to the displacement of the impact plate in real time, without relying on additional force sensors, greatly simplifying the system hardware requirements. By constructing an energy estimation model, the entire compression-rebound process is intelligently divided into five key stages: contact, linear compression, nonlinear compression, rebound initiation, and complete rebound. The energy loss rate is calculated and quantified for each stage, achieving accurate assessment of material energy flow, thereby improving the accuracy of judging compression-rebound hysteresis or compression-rebound asymmetry anomalies. Specifically, step S410 includes:

[0108] S411: Calculate the kinetic energy of the impact plate. and potential energy :

[0109] m is the mass of the impact plate. It is the derivative of the displacement s(t) of the impact plate with respect to time t;

[0110] g is the acceleration due to gravity, and h(t) is the velocity of the impact plate relative to s. ref The height difference between the positions, s(t) is the instantaneous displacement of the impact plate at any time t, s ref This is the position of the polyurethane buffer when it is uncompressed and at rest.

[0111] S412: Calculation of Energy Dissipation Rate Defined as the negative derivative of total energy (the sum of kinetic and potential energy) with respect to time. Calculate the energy dissipation rate. This indicates that the system energy is being dissipated. Based on the energy dissipation rate, the entire compression-rebound cycle is divided into five stages: contact period (the impact plate begins to contact the buffer), linear compression period (the material exhibits linear elastic compression), nonlinear compression period (the material enters the plastic deformation stage), rebound initiation period (energy begins to be released, and rebound begins), and complete rebound period (the impact plate is close to returning to its initial position). This scheme intelligently divides the entire compression-rebound process into five stages based on the variation characteristics of Φ(t) and its rate of change.

[0112] S413: Switching thresholds for five stages , It is the set of key time points for threshold determination, Δ (Φi) Based on normal sample statistical settings It is a set of segmented switching time points for threshold determination, wherein the time point when the threshold is first reached or exceeded within a stage is taken as the segmented switching time point; These correspond to the switching times of the four stages: the contact period, the linear compression period, the nonlinear compression period, and the rebound initiation period. There is an endpoint τ5 within the entire compression and rebound cycle, which is the moment when video recording ends in step S1.

[0113] Energy loss rate Λ at each stage k The calculation is as follows: k=1, hour, It is the starting moment. This refers to the moment when video recording begins in step S1; k=5. hour, It's the finish line, the finish line. This refers to the moment when video recording ends in step S1.

[0114] Energy loss rate reflects the changes in energy loss of polyurethane materials at each stage, and can sensitively reveal the differences in energy loss due to abnormal energy accumulation and asymmetric rebound during rebound hysteresis.

[0115] Furthermore, this invention employs a method combining time-domain and frequency-domain analysis to calculate the change in damping ratio of the polyurethane material during oscillation in real time. This dynamic damping ratio index accurately reflects the evolution of the viscoelastic properties of the polyurethane material over time, serving as a powerful aid to energy dissipation rate and improving the sensitivity and accuracy of anomaly detection. Specifically, step S420, for each stage divided in step S410, utilizes historical displacement data and combines time-domain and frequency-domain analysis methods to estimate the damping ratio in parallel, thus measuring the material's oscillation attenuation capability and revealing its viscoelasticity. Step S420 includes:

[0116] S421: Obtain the displacement peak values ​​for each stage from the continuous displacement curve. These displacement peak values ​​constitute a sequence of displacement oscillation peak values ​​within each stage. Calculate the displacement oscillation decrease in each stage. : , It is the number of displacement peaks;

[0117] S422: Calculate the damping ratio using the time-domain method at each stage: , p The logarithmic decay rate is used; then the damping ratio is calculated using frequency domain analysis. The dominant frequency fr and half-power bandwidth Δfr in the entire continuous displacement curve are extracted by Fourier transform.

[0118] The stage dynamic damping ratio is obtained by weighted fusion. , ，c t , c f It's weight. c t , c f ∈[0,1].

[0119] It should be noted that when the elevator car's impact plate strikes the polyurethane buffer, the buffer is compressed, causing deformation of the internal material and storing energy. Upon release, the polyurethane material rebounds, but due to internal friction and viscous energy dissipation, it cannot return to a stationary state immediately. Instead, it undergoes a process of "overshoot – repeated vibration – gradual subsidence." This "aftershock" is material oscillation. A sensor combination of an accelerometer and a data acquisition card is used to acquire time-domain waveforms and perform frequency-domain analysis. c t This represents whether the time-domain signal is clean and regular. If the attenuation curve is normal, then the time-domain result is given greater weight. c f This represents the clarity of the frequency domain peaks; if the formants are very sharp, the frequency domain result receives greater weight. Both are dynamically adjusted based on the characteristics of the real-time signal. Regarding the value range, Actual engineering values ​​are often in . It can be dynamically adjusted based on the quality of the signal transmitted from the sensor.

[0120] The dynamic damping ratio in step S420 reflects the damping characteristics of material oscillation and is the manifestation of "vibration decay" of polyurethane material at each stage. It can reveal the changes in viscoelastic properties caused by hysteresis or asymmetry.

[0121] In step S430 of this invention, the comparison is based solely on the correspondence of energy progress. For example, during compression, the energy is consumed to 50%; during rebound, the energy is released to 50%. The velocities at these two moments are compared to determine whether the speed and behavior are symmetrical. This places the compression and rebound processes on the same "energy scale," eliminating the superficial differences in time and speed, and more intuitively revealing the lag and asymmetry in energy absorption and release of the buffer. Specifically, the speed curves of compression and rebound are compared. A weighted integral form is used to construct a coupling comparison function to evaluate the dynamic correlation between compression and rebound, with a similarity score S. CR(i.e., coupling comparison) The closer the result is to 1, the more consistent the dynamic behavior of compression and rebound; a significantly lower result indicates the presence of hysteresis or asymmetry. Specifically, step S430 includes:

[0122] S431: Calculate the compression rate curve during the linear compression period. , S c2 It is the displacement during the linear compression period obtained based on the continuous displacement curve;

[0123] Calculate the compression rate curve during the nonlinear compression period. , S c3 It is the displacement during the nonlinear compression period obtained based on the continuous displacement curve;

[0124] Calculate the rebound speed curve during the rebound initiation phase. , S r4 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve;

[0125] Calculate the rebound speed curve during the full rebound period. , S r5 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve;

[0126] S432: Based on the cosine similarity principle, design a coupling comparison function S for the compression velocity curve and the rebound velocity curve. CR : S CR The value range is [0, 1], S CR The closer the value is to 1, the more consistent the shape and phase of the compression velocity curve and the rebound velocity curve are. for or , for or ;

[0127] Based on the coupling comparison function S CR ,calculate and Coupling comparison function S CR1 and calculation and Coupling comparison function S CR2 .

[0128] Based on the cosine similarity principle, this function is constructed by unifying the directions of compression and rebound velocities. The numerator of this function is equivalent to a dot product, used to measure the degree of synchronous change of the two curves over the entire time interval, while the denominator is equivalent to a product of moduli, eliminating the influence of curve amplitude differences through normalization. The resulting S... CR The value range is [0, 1]. The closer the value is to 1, the more consistent the compression and rebound speed curves are in shape and phase. Step S430 can set an early warning / trigger that only undertakes alarm and positioning responsibilities. The final conclusion is based on the evaluation model E.

[0129] Reference Figure 2 Furthermore, the method for handling the video of the impact plate and polyurethane buffer in step S1 includes:

[0130] When the displacement sensor measures that the physical displacement of the impact plate is less than the set threshold for the first time and the duration exceeds the threshold frame length, it is determined that the impact plate is about to make physical contact with the polyurethane buffer. This moment is marked as the starting point of the compression process, and video recording is triggered.

[0131] When the displacement sensor measures that the physical displacement of the impact plate is greater than the set threshold and the duration exceeds the set time, the video recording ends and the video is saved.

[0132] Step S2 obtains the pixel-level displacement of the impact plate in the recorded video.

[0133] This solution uses physical displacement data to control the start and stop of video recording, significantly reducing data processing volume and storage space requirements. In this invention, video recording and compression status detection form a complete closed loop: simultaneously recording the compression process, tracking the target, and performing status comparison analysis ensures data integrity and traceability, meeting the current elevator inspection regulations' requirements for the objectivity and traceability of test records.

[0134] In one embodiment of the present invention, the method for preprocessing the image frame in step S1 includes:

[0135] S101: Lens Distortion Correction: Based on the camera intrinsic parameter matrix and distortion coefficients, geometric correction is performed using OpenCV's undistort(·) function to eliminate radial and tangential distortions caused by wide-angle lenses, ensuring that the shape and position of the buffer region are accurately reproduced in the image. The distortion coefficients of OpenCV's undistort(·) function were obtained through camera calibration experiments.

[0136] S102: Image Contrast Enhancement: Convert the image to the LAB color space and perform adaptive histogram equalization (CLAHE) on the L luminance channel to enhance local image contrast and edge features, thereby improving image quality and detail recognition under complex lighting conditions.

[0137] S103: Region of Interest Extraction: Based on the prior position of the buffer in the image or the result of dynamic target localization, the central region or a specified region of the image is cropped as the Region of Interest (ROI) to remove redundant background information at the image edges and focus the analysis on the target region.

[0138] S104: Image Size Normalization: The cropped image frames are scaled to a set pixel size to meet the input size requirements of the deep learning object detection model and improve the model's inference speed and the processing efficiency of edge devices. In this embodiment of the invention, the cropped image frames are uniformly scaled to 640×640 pixels.

[0139] This solution constructs an image preprocessing pipeline through steps S101-S104.

[0140] Accordingly, the present invention also provides an elevator polyurethane buffer compression performance testing system, which is used to implement the above-mentioned elevator polyurethane buffer compression performance testing method.

[0141] The system includes a housing and a camera, displacement sensor, edge module, touch screen and communication module mounted on the housing. The YOLOv8 network model and the DeepSORT algorithm are deployed on the edge module.

[0142] The camera is used to record video of the impact plate and the polyurethane buffer during the compression and rebound processes of the polyurethane buffer.

[0143] The displacement sensor is used to measure the physical displacement of the impact plate in real time;

[0144] The edge module is used to preprocess image frames, obtain pixel-level displacement of the impact plate, and determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly and generate a detection report.

[0145] The touchscreen is used to play videos and to input device information;

[0146] The communication module is used to upload the test report to the cloud platform or server.

[0147] The system of this invention features high integration; the camera, displacement sensor, edge module, touchscreen, and communication module can be combined into a small housing, making it portable and adaptable to the space constraints and real-time requirements of elevator on-site inspection environments. Specifically, refer to... Figure 3 The edge module and communication module are housed within the casing 1. The touchscreen 2 is mounted on the top of the casing 1. The camera 3 is mounted on the side of the casing 1 via a telescopic support rod 4. The displacement sensor is mounted on the side of the camera 3 or on the top of the casing 1. When the camera 3 is in its retracted state (… Figure 4 The telescopic support rod 4 rotates to be close to the side wall of the outer casing 1, and the camera 3 is located inside the storage compartment (not shown in the figure); when the camera 3 is in working condition ( Figure 3 The telescopic support rod 4 supports the camera located above the housing 1. During testing, the system of this invention is placed in the elevator pit, and the position of the camera 3 is adjusted so that the lens is aimed at the polyurethane buffer; the system equipment is started, the system is first initialized, and then video recording is triggered and terminated based on the physical displacement of the impact plate measured by the displacement sensor.

[0148] Specifically, the system of this invention employs an edge computing development board, i.e., an edge module. This board carries an NPU computing module and is configured with a GPU / CPU / NPU task scheduling strategy to achieve efficient utilization of computing resources. During system initialization, the industrial camera is initialized, and automatic exposure, automatic gain, and image distortion correction parameters are configured to ensure clear and stable video images in complex lighting environments such as elevator pits. Simultaneously, a synchronous acquisition thread is started to provide raw visual input for target detection and status recognition. This embodiment of the invention uses a Sony IMX662 high frame rate industrial camera with a resolution of 1080P and a frame rate of up to 120fps. The laser rangefinder sensor initialization calibration process includes zero-point offset compensation, temperature drift correction, and multi-region depth calibration to ensure real-time measurement of the polyurethane buffer compression displacement within an accuracy range of ±3mm. This embodiment of the invention uses an STMicroelectronics VL53L5CX multi-channel Time-of-Flight (ToF) laser rangefinder sensor.

[0149] Other components and operations of the elevator polyurethane buffer compression performance testing method and system according to embodiments of the present invention are known to those skilled in the art and will not be described in detail here.

[0150] In the description of this specification, references to terms such as "embodiment," "example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0151] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for testing the compressibility of an elevator polyurethane buffer, characterized in that, This method is applied to an elevator polyurethane buffer compression performance testing system. The system includes a camera, a displacement sensor, an edge module, a touch screen, and a communication module. The edge module is equipped with a YOLOv8 network model and a DeepSORT algorithm, and the touch screen is used to input device information. The method includes the following steps: S1: The camera acquires video of the collision plate and the polyurethane buffer during the compression and rebound processes of the polyurethane buffer, and the edge module preprocesses the image frames. S2: Input the preprocessed image frames into the trained YOLOv8 network model to perform target recognition on the images; based on the target recognition results, perform inter-frame target tracking using the DeepSORT algorithm to obtain the pixel-level displacement of the impact plate; measure the physical displacement of the impact plate in real time using a displacement sensor; S3: The pixel-level displacement and physical displacement are fused to construct a continuous displacement curve of the compression rebound of the polyurethane buffer; S4: Based on the continuous displacement curve of the compression rebound of the polyurethane buffer, calculate the energy loss rate of the impact plate and the dynamic damping ratio of the polyurethane buffer, and construct the coupling comparison function of compression rebound; based on the fusion index of energy loss rate and dynamic damping ratio and the coupling comparison function of compression rebound, determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly. S5: Generate an inspection report for the elevator polyurethane buffer based on the continuous displacement curve, anomaly judgment results, and equipment information; S6: The communication module uploads the test report to the cloud platform or server; In step S2, dynamic compensation for environmental factors based on material properties is performed on the measured physical displacement, including: S201: Establish a nonlinear temperature compensation model: ; Among them, D a It is the physical displacement after temperature compensation, D raw These are the measured physical displacements, and α1, α2, and α3 are the first-, second-, and third-order temperature drift coefficients, respectively. It is the temperature difference coefficient; K material () represents the temperature-dependent modulus correction factor for polyurethane materials, ζ1, ζ2, and ζ3 are aging degradation coefficients, N is the cumulative compression cycles, and N0 is the rated life cycles. It is the cumulative frequency coefficient. , It uses a time coefficient. t is the elapsed time, and t0 is the design life. It is the load application factor. , It is the stress corresponding to the rated load. It is the first Constant working stress, It is the cumulative stress-time integral; S202: Establish a humidity-temperature coupled compensation model: , where D b This represents the physical displacement after temperature and humidity coupling compensation, β1 and β2 are the first and second order influence coefficients of humidity, ρ is the temperature and humidity interaction coefficient, and H is the humidity coefficient. , This is the reference calibration humidity value. It refers to relative humidity; S203: Calculate the compensated physical displacement: .

2. The method according to claim 1, characterized in that, Step S3 includes: S301: Convert pixel-level displacement to physical displacement: , where L vision (t) represents the calculated physical displacement. This represents the target motion quantity in the pixel domain. This represents the mapping coefficient from pixel to physical displacement obtained from system calibration; S302: Interpolation is used to align physical displacement with pixel-level displacement in time. , where L tof (t s ) represents the time t to be aligned. s The physical displacement, t a and t b Distance t s The two most recent sampling times, L tof (t a ) is t a Physical displacement at the sampling time, L tof (t b ) is t b Physical displacement at the sampling time; The calculated physical displacement and the time-aligned physical displacement are merged to generate a continuous displacement curve for the compression rebound of the polyurethane buffer: Where λ1 and λ2 are adaptive weighting coefficients, satisfying λ1+λ2=1; S303: The fused data is processed using the Kalman filter algorithm to reconstruct the smooth trajectory of the continuous displacement curve.

3. The method according to claim 1, characterized in that, Step S4 includes: S410: Construct an energy estimation model, combining displacement data from the continuous displacement curve, and divide the entire compression and rebound cycle of the polyurethane buffer into five stages: contact period, linear compression period, nonlinear compression period, rebound initiation period, and full rebound period. Calculate the energy loss rate of the impact plate in each stage. ; S420: Based on the stages defined in step S410, calculate the dynamic damping ratio for each stage. ; S430: Based on the stages divided in step S410, calculate the compression rate curves for the linear compression period and the nonlinear compression period, and calculate the rebound rate curves for the rebound initiation period and the full rebound period, respectively. Construct a coupling comparison function S for the compression rate curve and the rebound rate curve. CR ; Calculate the coupling comparison function S of the compression velocity curve during the linear compression period and the rebound velocity curve during the full rebound period. CR1 And the coupled comparison function S for calculating the compression velocity curve during the nonlinear compression period and the rebound velocity curve during the rebound initiation period. CR2 ; S440: Constructing the evaluation model: ; Where E is the overall score, and w1, w2, and w3 are the weights. The average energy loss rate represents the normal sample. The standard deviation of the energy loss rate represents the normal sample. The mean dynamic damping ratio represents the normal sample. The standard deviation of the dynamic damping ratio represents the normal sample. The mean coupling score representing the normal sample. The standard deviation of the coupling score represents the normal sample; when calculating the comprehensive score E during the contact period, let... When calculating the combined score E for the linear compression period and the full rebound period, let S... CR =S CR1 When calculating the combined score E for the nonlinear compression period and the rebound initiation period, let S... CR =S CR2 ; The overall score E for each stage is compared with the corresponding threshold. If the overall score E for any stage exceeds the corresponding threshold... When this occurs, it is determined that there is a compression rebound lag or a compression rebound asymmetry anomaly.

4. The method according to claim 3, characterized in that, Step S410 includes: S411: Calculate the kinetic energy of the impact plate. and potential energy : m is the mass of the impact plate. It is the derivative of the displacement s(t) of the impact plate with respect to time t; g is the acceleration due to gravity, and h(t) is the velocity of the impact plate relative to s. ref The height difference between the positions, s(t) is the instantaneous displacement of the impact plate at any time t, s ref This is the position of the polyurethane buffer when it is uncompressed and at rest. S412: Calculate the energy dissipation rate , This indicates that the system energy is being dissipated. Based on the energy dissipation rate, the entire compression-rebound cycle is divided into five stages: the contact period, the linear compression period, the nonlinear compression period, the rebound initiation period, and the complete rebound period. S413: , It is a set of segmented switching time points for threshold determination, wherein the time point when the threshold is first reached or exceeded within a stage is taken as the segmented switching time point; Energy loss rate Λ at each stage k The calculation is as follows: k=1, hour, It is the starting moment. This refers to the moment when video recording begins in step S1; k=5. hour, It's the finish line, the finish line. This is the moment when video recording ends in step S1.

5. The method according to claim 4, characterized in that, Step S420 includes: S421: Obtain the displacement peak values ​​for each stage from the continuous displacement curve. These displacement peak values ​​constitute a sequence of displacement oscillation peak values ​​within each stage. Calculate the displacement oscillation decrease in each stage. : , It is the number of displacement peaks; S422: Calculate the damping ratio using the time-domain method at each stage: , p The logarithmic decay rate is used; then the damping ratio is calculated using frequency domain analysis. The dominant frequency fr and half-power bandwidth Δfr in the entire continuous displacement curve are extracted by Fourier transform. The stage dynamic damping ratio is obtained by weighted fusion. , ，c t , c f It's weight. c t , c f ∈[0,1].

6. The method according to claim 4, characterized in that, Step S430 includes: S431: Calculate the compression rate curve during the linear compression period. , S c2 It is the displacement during the linear compression period obtained based on the continuous displacement curve; Calculate the compression rate curve during the nonlinear compression period. , S c3 It is the displacement during the nonlinear compression period obtained based on the continuous displacement curve; Calculate the rebound speed curve during the rebound initiation phase. , S r4 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve; Calculate the rebound speed curve during the full rebound period. , S r5 It is the displacement during the rebound initiation period obtained based on the continuous displacement curve; S432: Based on the cosine similarity principle, design a coupling comparison function S for the compression velocity curve and the rebound velocity curve. CR : S CR The value range is [0, 1], S CR The closer the value is to 1, the more consistent the shape and phase of the compression velocity curve and the rebound velocity curve are. for or , for or ; Based on the coupling comparison function S CR ,calculate and Coupling comparison function S CR1 and calculation and Coupling comparison function S CR2 .

7. The method according to claim 1, characterized in that, The method for handling the video of the impact plate and polyurethane buffer in step S1 includes: When the displacement sensor measures that the physical displacement of the impact plate is less than the set threshold for the first time and the duration exceeds the threshold frame length, it is determined that the impact plate is about to make physical contact with the polyurethane buffer. The moment when it is determined that the impact plate is about to make physical contact with the polyurethane buffer is marked as the starting point of the compression process, and video recording is triggered. When the displacement sensor measures that the physical displacement of the impact plate is greater than the set threshold and the duration exceeds the set time, the video recording ends and the video is saved. Step S2 obtains the pixel-level displacement of the impact plate in the recorded video.

8. The method according to claim 7, characterized in that, In step S1, the method for preprocessing the image frame includes: S101: Lens distortion correction: Based on the camera intrinsic parameter matrix and distortion coefficients, geometric correction is performed using OpenCV's undistort() function to eliminate radial and tangential distortion caused by wide-angle lenses; S102: Image contrast enhancement: Convert the image to the LAB color space, perform adaptive histogram equalization on the brightness channel, and enhance the local contrast and edge features of the image. S103: Region of Interest Extraction: Based on the prior position of the buffer in the image or the result of dynamic target localization, crop the central region of the image or a specified region as the region of interest; S104: Image size normalization: Scales the cropped image frame to the set pixel size.

9. A system for testing the compressibility of an elevator polyurethane buffer, characterized in that, This system is used to implement the elevator polyurethane buffer compression performance testing method according to any one of claims 1-8; The system includes a housing and a camera, displacement sensor, edge module, touch screen and communication module mounted on the housing. The YOLOv8 network model and the DeepSORT algorithm are deployed on the edge module. The camera is used to record video of the impact plate and the polyurethane buffer during the compression and rebound processes of the polyurethane buffer. The displacement sensor is used to measure the physical displacement of the impact plate in real time; The edge module is used to preprocess image frames, obtain pixel-level displacement of the impact plate, and determine whether the polyurethane buffer has compression rebound hysteresis or compression rebound asymmetry anomaly and generate a detection report. The touchscreen is used to play videos and to input device information; The communication module is used to upload the test report to the cloud platform or server.

Images