Sliding block meshing depth detection method based on visual technology

Abstract

The invention discloses a sliding block meshing depth detection method based on a visual technology, and belongs to the technical field of elevator detection. The method comprises the steps that firstly, historical image records and configuration data of an elevator are collected, a static image of an elevator door sliding block is collected through an optical imaging system, a virtual space is established according to the configuration data, and an elevator door virtual movement simulation demonstration scene is constructed; secondly, analyzing historical image records by adopting a visual technology, and establishing a typical speed change curve; according to the relation between the position and the speed in the typical speed change curve, a virtual physical engine is adopted in the virtual space to invert and generate a meshing scheme; then, the matching index of each meshing scheme is calculated in combination with static images collected in real time, alternative schemes are set, the abnormal index is calculated according to all the alternative schemes, and an abnormal sliding block is dynamically set; and finally, the relative position and the meshing depth of the abnormal sliding block are prompted in an early warning mode through a visual screen, and therefore a maintainer is assisted in completing elevator sliding block meshing maintenance work.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elevator detection, in particular to a sliding block engagement depth detection method based on visual technology. BACKGROUND

[0002] In the modern elevator safety operation system, the door system sliding block as the core transmission component, its engagement depth directly affects the stability of the door operation and the reliability of the emergency locking. Such components are subjected to high frequency opening and closing friction for a long time, and are prone to progressive wear on the hidden contact surface. Traditional monitoring methods are difficult to penetrate complex coupled structures, and cannot evaluate the deep mechanical engagement state in real time, resulting in hidden faults that are often in an unknowable black box state.

[0003] At present, traditional detection methods rely on regular manual inspection or local image sampling, which is difficult to detect the internal engagement state, and cannot capture the dynamics abnormalities caused by wear during movement. This structural detection blind area forces equipment maintenance to adopt a passive response mode, not only increasing the risk of sudden derailment, but also increasing the life cycle operation and maintenance cost. For example, 1. Visual analysis stays at the level of surface texture recognition, which can only capture surface scratches or deformation, and lacks a quantitative model for the degradation of deep engagement interface geometric features. 2. The detection data presents an island feature, and the speed curve collected by the motion sensor and the shape parameters identified by the vision cannot establish a coupled analysis mechanism, resulting in the breakdown of the causal relationship between material wear and dynamic response. 3. The repair decision seriously depends on fixed threshold rules, which cannot correct the error boundary by integrating motor drive characteristics, material properties and other multi-source physical parameters, nor can it predict the implicit wear evolution path through virtual deduction. This static and fragmented detection logic is very easy to cause fault early warning lag and vicious cycle of maintenance resource mismatch. SUMMARY

[0004] The purpose of the present application is to provide a sliding block engagement depth detection method based on visual technology to solve the problems raised in the background art.

[0005] In order to solve the above technical problems, the present application provides a sliding block engagement depth detection method based on visual technology, comprising:

[0006] S100, collect historical image records and configuration data of the elevator, collect static images of the elevator door sliding block through an optical imaging system, establish a virtual space according to the configuration data, and construct an elevator door virtual movement simulation demonstration scene.

[0007] The historical image record refers to the video picture collected inside the elevator by the camera.

[0008] The configuration data includes model data, power data and material data.

[0009] Model data refers to the three-dimensional models of each component in the elevator equipment, including geometric size shape and relative spatial position.

[0010] Geometric size shape refers to the physical profile and dimensional data of the component in Euclidean space. Relative spatial position refers to the assembly constraints and motion relationships between components.

[0011] Power data refers to motor drive parameters and electrical control parameters during the movement of the elevator door.

[0012] Motor drive parameters include at least: reduction ratio, used for output torque amplification. Screw lead, door displacement per motor revolution. Synchronous pulley ratio, pulley diameter ratio determines the speed ratio. Mechanism transmission efficiency, energy transmission loss rate. Moment of inertia, system resistance to speed change ability. Connecting rod swing angle, crank slider mechanism motion range.

[0013] Electrical control parameters include at least: bus voltage, inverter input reference. PWM carrier frequency, determines the motor current ripple. Phase current limiting, protects the motor from overload.

[0014] Motor drive parameters are used to describe the quantitative characteristics of the mechanical energy transmission path from the motor to the door, and electrical control parameters are used to control the strategy and boundary conditions of the mechanical energy conversion process.

[0015] Motor drive parameters determine the physical capability boundary of the door machine system, while electrical control parameters realize precise motion trajectory shaping within this boundary. Both are cooperatively modeled in digital twin simulation.

[0016] Material data includes the physical properties of each component in the elevator equipment.

[0017] Physical properties refer to the inherent characteristics exhibited by materials under the action of physical fields, which directly determine the response behavior of components in mechanical, thermal, and electrical environments.

[0018] Physical properties include at least mass, yield strength, elastic modulus, thermal expansion coefficient, thermal conductivity, specific heat capacity, resistivity, magnetic permeability, and dielectric strength.

[0019] Building an elevator door virtual movement simulation demonstration scene includes:

[0020] S101, using digital twin technology to establish a virtual space, analyzing the parameter set of the geometric size shape of each component in the model data, and describing the spatial transformation matrix of the assembly constraints.

[0021] S102, input the parameter set as model outline data into the three-dimensional modeling engine, and perform coordinate transformation operation according to the spatial transformation matrix, generate a three-dimensional elevator model containing each component model in the virtual space.

[0022] S103, According to the morphological characteristics of the elevator door slider in the static image collected by the optical imaging system, the corresponding contour parameters in the three-dimensional elevator model are adjusted in real time.

[0023] S104, Analyze the physical properties of each component in the material data, associate each physical property with the corresponding component in the three-dimensional elevator model, and generate a material card.

[0024] S105, Read the motor drive parameters and electrical control parameters in the power data, convert the motor drive parameters into translation mechanism motion relationship. Synchronously load the electrical control parameters, establish the elevator door motion control mode for the three-dimensional elevator model.

[0025] S106, Through the virtual physical engine combined with contour data, space transformation matrix, material card and motion control mode, a virtual moving simulation demonstration scene with interactive response characteristics is constructed in the virtual space.

[0026] Establish the basic virtual environment of elevator door motion, provide a benchmark platform for subsequent analysis. Convert real-world elevator historical image records and various configuration data into digital models, avoiding the complexity of direct experiments on physical devices.

[0027] Use digital twin technology to create a dynamic simulation environment to simulate the movement of the elevator door under different conditions. Real-time adjustment of the model through optical imaging collected static images ensures that the virtual environment is highly consistent with the real device, improving the reliability of subsequent simulation and detection.

[0028] S200, Analyze historical image records using visual technology and establish typical speed change curves. According to the position and speed relationship in the typical speed change curve, a virtual physical engine is used to generate a meshing scheme in the virtual space. Specifically including:

[0029] S201, Analyze the time sequence of historical image records, extract video data within the past time . Detect the target area where the elevator door is located in the video, and take the time period of continuous movement of the elevator door as the running period.

[0030] All running periods are within the past time , The setting of each running period needs to ensure that the elevator door runs normally and there is no stagnation caused by external interference.

[0031] S202, Draw a speed curve for each running period, and all speed curves are normalized based on the single motion distance of the elevator door. Align the time axis of all speed curves in the same category through dynamic time warping algorithm. Specifically including:

[0032] ​S2021、According to the running period, the video data is divided into different video segments , and the video segments of the running period are analyzed frame by frame, and the pixel position changes of the door feature marks in the continuous frames are identified by the visual template matching technology.

[0033] The visual template matching technology refers to a computer vision technology for searching similar areas in video frames through a predefined image template.

[0034] S2022, the visual template matching adopts a normalized cross-correlation algorithm to calculate the similarity of the target area, and selects the peak value position of the similarity as the center coordinates of the feature mark. The pixel displacement is converted into physical displacement to generate time and displacement data sequences.

[0035] The conversion of pixel displacement to physical displacement requires conversion calculation according to the camera calibration parameters set in advance, that is, setting the corresponding relationship parameters of pixel displacement length and real physical size.

[0036] S2023, numerical differentiation operation is performed on the time and displacement data sequence, and five-point central difference method is used to calculate the instantaneous speed. Draw the speed curve in the single moving process, and draw the speed curves of other running periods respectively.

[0037] The five-point central difference method is a numerical differentiation algorithm that uses five adjacent data points to calculate the derivative value to suppress measurement noise.

[0038] S203, a curve fitting method based on probability density estimation is used to calculate the speed value distribution function at each time point, and the point with the maximum probability density is taken as the speed value of the fitting curve at that time, to generate a typical speed change curve.

[0039] S204, analyze the relationship between the elevator door stop position and the instantaneous speed in the typical speed change curve, and use a virtual physics engine in a virtual space to generate a meshing scheme according to the instantaneous speed at different elevator door stop positions. Specifically includes:

[0040] S2041, analyze the profile parameters of each elevator door slider in the three-dimensional elevator model, analyze the known profile parameters of the corresponding elevator door slider according to the morphological characteristics presented in the static image, and the morphological characteristics not presented are taken as unknown profile parameters.

[0041] S2042, obtain the typical speed change curve, analyze the instantaneous speed corresponding to each elevator door stop position with the time axis as the reference . Set the allowable fluctuation threshold , calculate the speed interval corresponding to each stop position .

[0042] Among them, ​is represented as ;

[0043] The dwell position refers to the static point position of the elevator door in the stroke relative to the fixed reference system of the shaft, and the instantaneous speed refers to the instantaneous value of the speed of the elevator door before reaching the static point along the track movement.

[0044] S2043, a mapping relationship between the position and the speed interval is established, and a trigger condition is set according to the mapping relationship. The wear degree interval is set, so that different prediction values are respectively preset for each unknown profile parameter of the elevator door slider.

[0045] The wear degree interval is set according to the reasonable wear interval of the elevator operation data, the physical properties of the slider, and the material. The setting of the prediction value also needs to refer to the profile parameters of the slider after wear corresponding to different wear degrees in the wear degree interval.

[0046] S2044, a demonstration scene is established according to the measured value of all known profile parameters and the prediction value of unknown profile parameters. In the virtual space, a single virtual movement simulation inversion is performed on the elevator door in the three-dimensional elevator model through a virtual physical engine.

[0047] S2045, after the inversion is completed, the relationship between the position and the speed is analyzed, the trigger condition is met, and the engagement scheme is established. The prediction value of all unknown profile parameters is put into the engagement scheme.

[0048] S2046, different demonstration scenes are continuously established for inversion until the number of engagement schemes meets the preset threshold, the inversion is stopped, and the prediction values of the unknown profile parameters in different engagement schemes are not completely the same.

[0049] In each demonstration scene, the measured value of all known profile parameters is fixed, and the prediction value of all unknown profile parameters is not fixed. In different demonstration scenes, the prediction values of all unknown profile parameters are not completely the same.

[0050] Key motion laws are extracted from historical data, and possible wear states of the slider are predicted. Typical speed change curves are generated by analyzing historical images, motion patterns of the elevator door are captured, and noise interference is reduced.

[0051] Based on the relationship between the position and the speed, a virtual physical engine is used to generate an engagement scheme, and a prediction framework is provided.

[0052] S300, during the maintenance process, the matching index of each engagement scheme is calculated respectively combined with the real-time collected static images, and the alternative scheme is set. The abnormal index of each elevator door slider is calculated according to all alternative schemes, and the abnormal slider is dynamically set. Specifically, it includes:

[0053] S301, analyze the engagement scheme According to the relationship between position and velocity in the corresponding inversion process, a velocity curve is drawn. According to the analysis of the measured value of the corresponding unknown profile parameter in the newly collected static image in the maintenance process, the matching index is calculated .

[0054] The matching index calculation formula is as follows:

[0055] ;

[0056] In the formula, is the weight, is the similarity between the velocity curve of the meshing scheme and the typical velocity change curve, is the number of unknown profile parameters in the newly collected static image that are newly added and exist in the meshing scheme .

[0057] is the predicted value of the th unknown profile parameter, is the measured value of the th unknown profile parameter.

[0058] The matching index measures the shape similarity between the velocity curve generated by the inversion of the meshing scheme and the typical velocity change curve. This similarity value constitutes the main base of the formula.

[0059] By introducing the weight-regulated deviation penalty term, the average relative error between the predicted values of all verifiable unknown profile parameters in the meshing scheme and the newly collected measured values is calculated.

[0060] The final result is jointly affected by the positive driving of the main similarity and the inverse inhibition of the predicted deviation.

[0061] When the velocity curve is highly matched and the predicted parameter deviation is small, the index tends to be the maximum. Conversely, if the prediction deviates significantly from the actual measurement or the velocity curve is distorted, the index is significantly reduced.

[0062] A dynamic screening mechanism is constructed to ensure that the selected meshing scheme meets the dual conditions of motion law reproduction and parameter inference reliability.

[0063] S302, calculate the matching index of each meshing scheme respectively, and select the meshing scheme with a matching index greater than a threshold value as a candidate scheme. Calculate the abnormal index of each elevator door slider according to all candidate schemes, and set the abnormal slider. Specifically, it includes:

[0064] S3021, modify the unknown profile parameter with a measured value to a known profile parameter, and delete it from all candidate schemes. Calculate the standard deviation of the predicted value of the same unknown profile parameter in all candidate schemes as its fluctuation coefficient.

[0065] S3022. Based on the geometric dimensions and shapes of each elevator door slider in the model data, analyze and obtain the initial values ​​of each contour parameter. Analyze the elevator door slider to which each contour parameter belongs, and substitute them into the formula to calculate the anomaly index. :

[0066] ;

[0067] In the formula, and These are the weights for the known contour parameters and the unknown contour parameters, respectively. and These represent the number of known and unknown contour parameters in the elevator door slider, respectively.

[0068] and The first Initial and measured values ​​of known contour parameters. For the first Initial values ​​for unknown contour parameters, For the first The unknown profile parameter is the average of the predicted values ​​across all alternatives.

[0069] For the first The fluctuation coefficient of an unknown profile parameter This represents the maximum fluctuation coefficient for all unknown profile parameters.

[0070] The anomaly index aims to provide a comprehensive quantitative assessment of the overall abnormal state of each elevator door slider.

[0071] Its calculations consist of two physically significant branches:

[0072] The first part focuses on known profile parameters and directly reflects the cumulative effect of physical wear by comparing the mean relative deviation between the measured values ​​and the initial values.

[0073] The second part addresses the indirect evaluation of unknown profile parameters that cannot yet be directly measured, by combining the average deviation and fluctuation characteristics of their predicted values.

[0074] The volatility characteristic is manifested in the degree of dispersion of the predicted values ​​from different scenarios, with parameters exhibiting greater dispersion being assigned higher weights for anomalies. The two branches balance the intensity of influence through independent weighting coefficients, and the anomaly indication role of highly uncertain parameters is strengthened by introducing a normalization operation based on the maximum volatility coefficient.

[0075] This index ultimately integrates directly observable visible wear with implicit losses that require inversion to form a three-dimensional diagnosis of the slider's health status, providing a mathematical basis for accurately locating high-risk components.

[0076] Combine the new static image to calculate the matching index, screen out the high fitting degree of the alternative scheme, and filter the unrealistic prediction. Based on the data collected in real time during the maintenance process, the state of the slider is evaluated and the abnormality is screened.

[0077] Locate the specific slider problem through the abnormality index, and reduce false positives. Use mathematical formulas to integrate measured values and predicted values, so that the maintenance recommendations are objective and quantifiable.

[0078] S400, send the abnormal slider information to the intelligent terminal carried by the maintenance personnel, and warn the relative position and engagement depth of the abnormal slider through the visual screen, so as to assist the maintenance personnel to complete the maintenance work of the elevator slider engagement. Specifically, it includes:

[0079] S401, transmit the abnormal slider information to the portable intelligent terminal held by the maintenance personnel through a wireless communication channel, and the intelligent terminal automatically constructs a visual platform according to the three-dimensional model data in the virtual space.

[0080] S402, extract the relative spatial position data of the abnormal slider in the model data, and generate the relative direction mapping relationship of the abnormal slider relative to the parts at the elevator door and the engagement depth data in combination with the contour parameters.

[0081] S403, highlight the abnormal slider position on the visual screen of the intelligent terminal, and display the warning information formed by the abnormality index and the engagement depth data description text.

[0082] The analysis results are transmitted to the maintenance personnel to provide intuitive intervention. Through wireless communication, the abnormal slider data is pushed to the intelligent terminal in time to realize rapid response.

[0083] The slider position and engagement depth are highlighted on the terminal screen, reducing the cognitive burden of the maintenance personnel. The specific problem is warned to prevent safety incidents caused by neglecting details.

[0084] Compared with the prior art, the beneficial effects achieved by the present application are:

[0085] Stealth state perspective ability: Break through the limitation that traditional visual detection cannot penetrate solid structure, realize cross-dimension mapping of hidden engagement surface wear morphology through virtual space dynamic inversion technology, and eliminate the detection blind area of mechanical black box.

[0086] Multi-source data fusion mechanism: Establish a coupling analysis framework of kinematics response and material performance, unify the modeling of historical running track, motor drive characteristics and physical property parameters, and build a complete analysis system of wear attribution chain.

[0087] Predictive diagnosis paradigm: based on dynamic probability fitting to generate different alternatives, through virtual deduction to quantify the evolution path of implicit wear and tear, to upgrade passive threshold alarm to active abnormality prediction, breaking through the control bottleneck of sudden failure.

[0088] Adaptive decision system: introduce abnormality indicator quantification model, fuse double evaluation mechanism of measured deviation and predicted volatility, establish diagnosis boundary standard with self-correction according to working conditions, improve false alarm suppression ability in complex noise environment.

[0089] Closed-loop operation and maintenance guidance system: through the visualization platform, the virtual diagnosis result is mapped to the terminal equipment in real time, forming the decision-making closed loop of abnormal positioning-depth analysis-repair guidance, reversing the traditional maintenance mode of trial and error based on experience. BRIEF DESCRIPTION OF DRAWINGS

[0090] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, which together with the embodiments of the present application, serve to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0091] Figure 1 is a flowchart of the slider engagement depth detection method based on visual technology of the present application. DETAILED DESCRIPTION

[0092] The technical solutions in the embodiments of the present application will be described in detail below, with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0093] Embodiment one: please refer to Figure 1 The present application provides a slider engagement depth detection method based on visual technology, comprising:

[0094] S100, collect the historical image record and configuration data of the elevator, collect the static image of the elevator door slider through the optical imaging system, establish a virtual space according to the configuration data, and construct an elevator door virtual moving simulation demonstration scene.

[0095] The historical image record refers to the video picture collected by the camera in the elevator.

[0096] The configuration data includes model data, power data and material data.

[0097] The model data refers to the three-dimensional model of each part of the elevator equipment, including geometric size shape and relative spatial position.

[0098] In the implementation process, the geometric size shape refers to the physical profile and dimensional data of the parts in Euclidean space. The relative spatial position refers to the assembly constraints and motion relationships between the parts.

[0099] The power data refers to the motor drive parameters and electrical control parameters during the movement of the elevator door.

[0100] In the implementation process, the motor drive parameters at least include: the reduction ratio for output torque amplification, the lead screw pitch, the door displacement per motor revolution, the synchronous pulley ratio, the pulley diameter ratio determining the speed ratio, the mechanism transmission efficiency, the energy transmission loss rate, the moment of inertia, the system resistance to speed change ability, and the connecting rod swing angle.

[0101] The electrical control parameters at least include: the bus voltage, the inverter input reference, the PWM carrier frequency, the phase current limiting, and the motor overload protection.

[0102] The motor drive parameters are used to describe the quantitative characteristics of the mechanical energy transmission path from the motor to the door, and the electrical control parameters are used to control the strategy and boundary conditions of the process of converting electrical energy into mechanical energy.

[0103] In the implementation process, the motor drive parameters determine the physical capability boundary of the door machine system (such as maximum thrust and speed limit), and the electrical control parameters realize precise motion trajectory shaping (such as smooth start and stop and collision avoidance) within this boundary. The error in the collaborative modeling of the two in digital twin simulation needs to be controlled within ±1.5% of the transmission parameters and ±0.8% of the control parameters, so as to achieve a working condition reproduction accuracy of >90%.

[0104] The material data includes the physical properties of each part in the elevator equipment.

[0105] The physical properties refer to the inherent characteristics of materials under the action of physical fields, which directly determine the response behavior of parts in mechanical, thermal, and electrical environments.

[0106] In the implementation process, the physical properties at least include mass, yield strength, elastic modulus, thermal expansion coefficient, thermal conductivity, specific heat capacity, resistivity, magnetic permeability, and dielectric strength.

[0107] Building a virtual moving simulation demonstration scene of the elevator door includes:

[0108] S101, a virtual space is established using digital twin technology, and the parameter set of the geometric size shape of each part in the model data and the spatial transformation matrix describing the assembly constraints are analyzed.

[0109] S102, input the parameter set as model contour data into the three-dimensional modeling engine, and perform coordinate transformation operation according to the space transformation matrix to generate a three-dimensional elevator model containing each component model in the virtual space.

[0110] S103, real-time adjust the corresponding contour parameters in the three-dimensional elevator model according to the morphological characteristics of the elevator door slider in the static image collected by the optical imaging system.

[0111] S104, analyze the physical properties of each component in the material data, associate each physical property with the corresponding component in the three-dimensional elevator model, and generate a material card.

[0112] S105, read the motor drive parameters and electrical control parameters in the power data, convert the motor drive parameters into translation mechanism motion relationship, and simultaneously load the electrical control parameters to establish an elevator door motion control mode for the three-dimensional elevator model.

[0113] S106, construct an elevator door virtual moving simulation demonstration scene with interactive response characteristics in the virtual space by combining contour data, space transformation matrix, material card and motion control mode through a virtual physics engine.

[0114] Establish a basic virtual environment for elevator door motion to provide a benchmark platform for subsequent analysis. Convert real-world elevator historical image records and various configuration data into digital models to avoid the tediousness of direct experiments on physical devices.

[0115] In the specific implementation process, a dynamic simulation environment is created using digital twin technology to simulate the movement of the elevator door under different conditions. The model is adjusted in real time through static images collected by optical imaging to ensure that the virtual environment is highly consistent with the real device, improving the reliability of subsequent simulation and detection.

[0116] S200, analyze historical image records using visual technology and establish a typical speed change curve. According to the position and speed relationship in the typical speed change curve, a virtual physics engine is used to generate a meshing scheme in the virtual space. Specifically including:

[0117] S201, analyze the time sequence of historical image records, and intercept video data within the past time . Detect the target area where the elevator door is located in the video image, and take the time period of continuous movement of the elevator door as the running period.

[0118] In the specific implementation process, all running periods are within the past time , and the setting of each running period needs to ensure that the elevator door runs normally within the time period and there is no stagnation caused by external interference.

[0119] ​S202, draw a speed curve for each operating period, all speed curves are normalized based on the single motion distance of the elevator door, and the time axis of all speed curves of the same type is aligned through dynamic time warping algorithm. Specifically, it includes:

[0120] S2021, according to the operating period, the video data is split into different video segments, and the video segments of the operating period are analyzed frame by frame, and the pixel position change of the door feature mark in the continuous frame is identified through visual template matching technology.

[0121] Visual template matching technology refers to a computer vision technology that searches for similar areas in video frames through predefined image templates.

[0122] S2022, visual template matching uses a normalized cross-correlation algorithm to calculate the similarity of the target area, and selects the peak value position of the similarity as the center coordinates of the feature mark. Convert pixel displacement to physical displacement to generate time and displacement data sequences.

[0123] In the specific implementation process, the conversion of pixel displacement to physical displacement needs to be converted and calculated according to the camera calibration parameters set in advance, that is, the corresponding relationship parameters of pixel displacement length and real physical size are set.

[0124] S2023, perform numerical differentiation operation on the time and displacement data sequence, and calculate the instantaneous speed using five-point central difference method. Draw the speed curve in the single movement process, and draw the speed curves of other operating periods respectively.

[0125] Five-point central difference method is a numerical differentiation algorithm that uses five adjacent data points to calculate the derivative value to suppress measurement noise.

[0126] S203, use a curve fitting method based on probability density estimation to calculate the speed value distribution function at each time point, take the point with the maximum probability density as the speed value of the fitted curve at that time, and generate a typical speed change curve.

[0127] S204, analyze the relationship between the elevator door stop position and the instantaneous speed in the typical speed change curve, and use a virtual physics engine in a virtual space to generate a meshing scheme according to the instantaneous speed at different elevator door stop positions. Specifically, it includes:

[0128] S2041, analyze the profile parameters of each elevator door slider in the three-dimensional elevator model, analyze the known profile parameters of the corresponding elevator door slider according to the morphological characteristics presented in the static image, and the morphological characteristics not presented are taken as unknown profile parameters.

[0129] S2042, obtain the typical speed change curve, analyze the instantaneous speed corresponding to each elevator door stop position based on the time axis Set the allowable fluctuation threshold Calculate the speed interval corresponding to each stop position respectively .

[0130] Wherein, Indicated as ;

[0131] In the specific implementation process, the stop position refers to the static point position of the elevator door relative to the fixed reference system of the shaft in the stroke, and the instantaneous speed refers to the instantaneous value of the speed of the elevator door before reaching the front track movement of the static point.

[0132] S2043, establish the mapping relationship between the position and the speed interval, and set the trigger condition according to the mapping relationship. Set the wear degree interval, so as to respectively preset different prediction values for each unknown profile parameter of the elevator door slider.

[0133] In the specific implementation process, the wear degree interval is set according to the reasonable wear interval of the elevator operation data, the physical properties of the slider, and the material. The setting of the prediction value also needs to refer to the profile parameters of the slider after wear corresponding to different wear degrees in the wear degree interval.

[0134] S2044, establish a demonstration scene according to the measured value of all known profile parameters and the prediction value of unknown profile parameters. In the virtual space, drive the elevator door in the three-dimensional elevator model to perform single virtual movement simulation inversion through the virtual physical engine.

[0135] S2045, after the inversion is completed, analyze the relationship between the position and the speed, and if the trigger condition is met, establish the engagement scheme, and put the prediction value of all unknown profile parameters into the engagement scheme.

[0136] S2046, continuously establish different demonstration scenes for inversion until the number of engagement schemes meets the preset threshold, then stop the inversion, and the prediction values of each unknown profile parameter in different engagement schemes are not completely the same.

[0137] In the specific implementation process, in each demonstration scene, the measured value of all known profile parameters is fixed, and the prediction value of all unknown profile parameters is not fixed. In different demonstration scenes, the prediction values of all unknown profile parameters are not completely the same.

[0138] Extract key motion rules from historical data and predict possible wear states of the slider. Analyze historical images (video records) to generate typical speed change curves, capture the motion mode of the elevator door, and reduce noise interference.

[0139] Based on the relationship between position and speed, use the virtual physical engine to generate the "engagement scheme" (i.e. the parameter combination of the slider in the ideal wear state), and provide a prediction framework.

[0140] S300, during the maintenance process, the matching index of each meshing scheme is calculated respectively combined with the real-time collected static images, and the alternative scheme is set. According to all the alternative schemes, the abnormal index of each elevator door slider is calculated, and the abnormal slider is dynamically set. Specifically, it includes:

[0141] S301, analyze the meshing scheme According to the relationship between position and speed in the inversion process, the speed curve is drawn. According to the analysis of the measured value of the unknown profile parameter corresponding to the newly collected static image in the maintenance process, the matching index is calculated .

[0142] The matching index calculation formula is as follows:

[0143] ;

[0144] In the formula, is the weight, is the similarity between the speed curve of the meshing scheme and the typical speed change curve, is the number of unknown profile parameters in the newly collected static image that are newly added and exist in the meshing scheme .

[0145] is the predicted value of the first unknown profile parameter, is the measured value of the first unknown profile parameter.

[0146] In the specific implementation process, the matching index measures the shape similarity between the speed curve generated by the inversion of the meshing scheme and the typical speed change curve, and the similarity value constitutes the main base of the formula.

[0147] By introducing the deviation penalty term of weight regulation, the average relative error between the predicted value of all verifiable unknown profile parameters in the meshing scheme and the newly collected measured value is calculated.

[0148] The final result is jointly affected by the positive driving of the main similarity and the reverse inhibition of the prediction deviation.

[0149] When the speed curve is highly matched and the prediction parameter deviation is small, the index tends to be the maximum. On the contrary, if the prediction deviates seriously from the actual measurement or the speed curve is distorted, the index is significantly reduced.

[0150] A dynamic screening mechanism is constructed to ensure that the meshing scheme selected as the alternative scheme meets the dual conditions of motion law reproduction and reliable parameter inference.

[0151] S302. Calculate the matching index for each meshing scheme, and select meshing schemes with matching indices greater than a threshold as candidate schemes. Calculate the anomaly index for each elevator door slider based on all candidate schemes, and set the anomaly slider. Specifically, this includes:

[0152] S3021. Modify unknown profile parameters with measured values ​​to known profile parameters and remove them from all alternatives. Calculate the standard deviation of the predicted values ​​of the same unknown profile parameter across all alternatives as its fluctuation coefficient.

[0153] S3022. Based on the geometric dimensions and shapes of each elevator door slider in the model data, analyze and obtain the initial values ​​of each contour parameter. Analyze the elevator door slider to which each contour parameter belongs, and substitute them into the formula to calculate the anomaly index. :

[0154] ;

[0155] In the formula, and These are the weights for the known contour parameters and the unknown contour parameters, respectively. and These represent the number of known and unknown contour parameters in the elevator door slider, respectively.

[0156] and The first Initial and measured values ​​of known contour parameters. For the first Initial values ​​for unknown contour parameters, For the first The unknown profile parameter is the average of the predicted values ​​across all alternatives.

[0157] For the first The fluctuation coefficient of an unknown profile parameter This represents the maximum fluctuation coefficient for all unknown profile parameters.

[0158] In practice, the anomaly index aims to comprehensively and quantitatively assess the overall abnormal state of each elevator door slider.

[0159] Its calculations consist of two physically significant branches:

[0160] The first part focuses on known profile parameters and directly reflects the cumulative effect of physical wear by comparing the mean relative deviation between the measured values ​​and the initial values.

[0161] The second part addresses the indirect evaluation of unknown profile parameters that cannot yet be directly measured, by combining the average deviation and fluctuation characteristics of their predicted values.

[0162] where the fluctuation characteristics represent the dispersion degree of each scheme prediction value, and the parameters with greater dispersion are given higher abnormal contribution weights. The two branches balance the influence strength through independent weight coefficients, and the abnormal indication of high uncertainty parameters is strengthened by introducing the normalization operation of the maximum fluctuation coefficient.

[0163] This index finally integrates the directly observable explicit wear and the implicitly inferred hidden loss, forming a stereoscopic diagnosis of the slider health status and providing a mathematical basis for accurately positioning high-risk components.

[0164] Combine the new static image to calculate the matching index, and select the high-fitting candidate scheme to filter unrealistic predictions. Based on the data collected in real time during the maintenance process, evaluate the slider state and select abnormalities.

[0165] Locate specific slider problems through the abnormal index to reduce false positives (such as only individual slider abnormalities rather than the entire system). Use mathematical formulas to integrate measured and predicted values to make maintenance recommendations objective and quantifiable.

[0166] S400, send the abnormal slider information to the smart terminal carried by the maintenance personnel, and through the visual screen, warn the relative position and engagement depth of the abnormal slider, thereby assisting the maintenance personnel to complete the maintenance work of the elevator slider engagement. Specifically, it includes:

[0167] S401, transmit the abnormal slider information to the portable smart terminal held by the maintenance personnel through a wireless communication channel, and the smart terminal automatically builds a visual platform based on the three-dimensional model data in the virtual space.

[0168] S402, extract the relative spatial position data of the abnormal slider within the model data, and generate the relative direction mapping relationship of the abnormal slider relative to the components at the elevator door and the engagement depth data in combination with the contour parameters.

[0169] S403, highlight the abnormal slider position on the visual screen of the smart terminal, and display the warning information formed by the abnormal index and the engagement depth data description text.

[0170] In the specific implementation process, the analysis results are transmitted to the maintenance personnel to provide intuitive intervention. Through wireless communication (such as 5G), abnormal slider data is pushed to the smart terminal (such as a mobile phone or tablet) in a timely manner to achieve rapid response.

[0171] Highlight the slider position and engagement depth on the terminal screen to reduce the cognitive burden of maintenance personnel. Early warning of specific problems (such as insufficient engagement risk) can prevent safety incidents caused by neglecting details.

[0172] It is to be noted that, in the present text, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0173] Finally, it should be noted that the above-mentioned only constitutes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications, equivalent replacements, improvements and the like of the technical solutions described in the foregoing embodiments can still be made. Any modifications, equivalent replacements, improvements and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A slider engagement depth detection method based on vision technology, characterized in that: The method includes: S100: Collect historical image records and configuration data of the elevator; acquire static images of the elevator door slider through an optical imaging system; establish a virtual space based on the configuration data; and construct a virtual movement simulation demonstration scene for the elevator door. S200: Analyze historical image records using visual technology and establish typical velocity change curves; based on the position-velocity relationship in the typical velocity change curves, use a virtual physics engine in virtual space to invert and generate meshing schemes. S300 During the maintenance process, the matching index of each meshing scheme is calculated based on the real-time acquired static images, and alternative schemes are set; the abnormal index of each elevator door slider is calculated based on all alternative schemes, and abnormal sliders are dynamically set. S400: The abnormal slider information is sent to the smart terminal carried by the maintenance personnel. The relative position and engagement depth of the abnormal slider are displayed on the visual screen, thereby assisting the maintenance personnel in completing the maintenance work of the elevator slider engagement.

2. The slider engagement depth detection method based on vision technology according to claim 1, characterized in that: In S100, historical image recording refers to video footage of the elevator interior captured by a camera; Configuration data includes model data, dynamic data, and material data; Model data refers to the three-dimensional models of each component in the elevator equipment, including geometric dimensions, shape, and relative spatial position; Geometric dimensions and shape refer to the physical outline and dimensional data of a component in Euclidean space; Relative spatial position refers to the assembly constraints and motion relationships between components; Power data refers to the motor transmission parameters and electrical control parameters during the movement of the elevator doors; Motor transmission parameters are used to describe the quantitative characteristics of the mechanical energy transfer path from the motor to the door leaf, while electrical control parameters are used to control the strategy and boundary conditions for the conversion of electrical energy into mechanical energy. Material data includes the physical properties of each component in the elevator equipment; Physical properties refer to the inherent characteristics of materials under the influence of physical fields. These characteristics directly determine the response behavior of components in mechanical, thermal, and electrical environments.

3. The slider engagement depth detection method based on vision technology according to claim 2, characterized in that: The construction of a virtual motion simulation demonstration scenario for elevator doors includes: S101. Use digital twin technology to establish a virtual space, analyze the parameter set of geometric dimensions and shapes of each component in the model data, and the spatial transformation matrix describing assembly constraints; S102. Input the parameter set as model contour data into the 3D modeling engine, and at the same time perform coordinate transformation operation according to the spatial transformation matrix to generate a 3D elevator model containing the models of each component in the virtual space. S103. Based on the shape characteristics of the elevator door slider in the static image acquired by the optical imaging system, adjust the corresponding contour parameters in the three-dimensional elevator model in real time. S104. Analyze the physical properties of each component in the material data, associate each physical property with the corresponding component in the 3D elevator model, and generate a material card. S105. Read the motor transmission parameters and electrical control parameters from the power data, and convert the motor transmission parameters into the motion relationship of the translation mechanism; synchronously load the electrical control parameters to establish the elevator door motion control mode for the three-dimensional elevator model. S106. By combining contour data, spatial transformation matrix, material cards and motion control modes with a virtual physics engine, a virtual motion simulation demonstration scene of an elevator door with responsive interactive characteristics is constructed in virtual space.

4. The slider engagement depth detection method based on vision technology according to claim 2, characterized in that: S200 includes: S201. Analyze the time series of historical image records and extract past durations. Video data within ; Detect the target area where the elevator door is located in the video frame, and take the time period during which the elevator door continues to move as the running period; S202. Draw speed curves for each running segment. All speed curves are normalized based on the single movement distance of the elevator door. The time axis of all speed curves in the same category is aligned by the dynamic time warping algorithm. S203. Using a curve fitting method based on probability density estimation, calculate the velocity value distribution function at each time point, take the point with the maximum probability density as the velocity value of the fitted curve at that moment, and generate a typical velocity change curve. S204. Analyze the relationship between the elevator door's stopping position and instantaneous speed in typical speed change curves. In virtual space, use a virtual physics engine to generate meshing schemes based on the instantaneous speed at different elevator door stopping positions.

5. The slider engagement depth detection method based on vision technology according to claim 4, characterized in that: S202 includes: S2021, Based on the runtime segment, video data Split into different video segments, for runtime segments The video clips were analyzed frame by frame, and the pixel position changes of the door feature marks in consecutive frames were identified through visual template matching technology. S2022: Visual template matching uses a normalized cross-correlation algorithm to calculate the similarity of the target region and selects the peak similarity position as the center coordinate of the feature marker; pixel displacement is converted into physical displacement to generate a time and displacement data sequence. S2023. Perform numerical differentiation on the time and displacement data sequence, and calculate the instantaneous velocity using the five-point central difference method; plot the velocity curve during a single movement, and plot the velocity curves for each of the other running segments.

6. The slider engagement depth detection method based on vision technology according to claim 4, characterized in that: S204 includes: S2041. Analyze the contour parameters of each elevator door slider in the 3D elevator model. Analyze the known contour parameters of the corresponding elevator door slider based on the morphological features presented in the static image. The morphological features that are not presented are taken as unknown contour parameters. S2042. Obtain typical speed change curves and analyze the instantaneous speed corresponding to each elevator door's stopping position based on the time axis. Set the allowable fluctuation threshold. Calculate the velocity range corresponding to each dwell position. ; in, Represented as ; S2043. Establish a mapping relationship between position and speed range, and set trigger conditions according to the mapping relationship; set wear degree range, so as to preset different predicted values ​​for each unknown contour parameter of each elevator door slider; S2044. Based on the measured values ​​of all known contour parameters and the predicted values ​​of unknown contour parameters, a demonstration scene is established. In the virtual space, the elevator door in the 3D elevator model is driven by the virtual physics engine to perform a single virtual movement simulation inversion. S2045. After the inversion is completed, analyze the relationship between position and velocity. If the triggering condition is met, establish a meshing scheme and put the predicted values ​​of all unknown profile parameters into the meshing scheme. S2046. Continuously establish different demonstration scenarios for inversion until the number of meshing schemes meets the preset threshold, then stop the inversion. The predicted values ​​of various unknown contour parameters in different meshing schemes are not exactly the same.

7. The slider engagement depth detection method based on vision technology according to claim 6, characterized in that: The S300 includes: S301, Analyze meshing scheme To correspond to the relationship between position and velocity during the inversion process, a velocity curve is plotted; based on the measured values ​​of the corresponding unknown contour parameters from newly acquired static images during maintenance, the matching index is calculated. ; S302. Calculate the matching index of each meshing scheme, and take the meshing scheme with the matching index greater than the threshold as the alternative scheme; calculate the abnormal index of each elevator door slider according to all alternative schemes, and set the abnormal slider.

8. The slider engagement depth detection method based on vision technology according to claim 7, characterized in that: In S301, the matching index is calculated using the following formula: ； In the formula, As weight, For meshing scheme The similarity between the velocity curve and the typical velocity change curve, Added to the newly acquired static images and in the meshing scheme The number of unknown contour parameters present in the sample; For the first Predicted values ​​of unknown contour parameters, For the first Measured values ​​of unknown profile parameters.

9. The slider engagement depth detection method based on vision technology according to claim 7, characterized in that: S302 includes: S3021. Modify unknown contour parameters with measured values ​​to known contour parameters and delete them from all alternative schemes; calculate the standard deviation of the predicted values ​​of the same unknown contour parameter in all alternative schemes as its fluctuation coefficient. S3022. Based on the geometric dimensions and shapes of each elevator door slider in the model data, analyze and obtain the initial values ​​of each contour parameter; analyze the elevator door slider to which each contour parameter belongs, and substitute them into the formula to calculate the anomaly index. : ； In the formula, and These are the weights for the known contour parameters and the unknown contour parameters, respectively. and These represent the number of known and unknown contour parameters in the elevator door slider, respectively. and The first Initial and measured values ​​of known contour parameters. For the first Initial values ​​for unknown contour parameters, For the first The average of the predicted values ​​of the unknown contour parameters across all alternatives; For the first The fluctuation coefficient of an unknown profile parameter This represents the maximum fluctuation coefficient for all unknown profile parameters.

10. The slider engagement depth detection method based on vision technology according to claim 7, characterized in that: The S400 includes: S401. The abnormal slider information is transmitted to the portable smart terminal held by the maintenance personnel through the wireless communication channel. The smart terminal automatically constructs a visualization platform based on the three-dimensional model data in the virtual space. S402. Extract the relative spatial position data of the abnormal slider from the model data, and combine the contour parameters to generate the relative direction mapping relationship and engagement depth data of the abnormal slider relative to the components at the elevator door. S403. Highlight the abnormal slider position on the visual screen of the smart terminal, and display the warning information formed by the abnormal index and engagement depth data description text.

Images