Automatically-controlled auxiliary inspection trolley for retarder

Abstract

The invention discloses an automatic control auxiliary inspection trolley for a retarder, belongs to the technical field of railway equipment inspection, and can solve the technical problems that the inspection efficiency of the retarder is low and the service life prediction is lack of accuracy. The trolley comprises a walking top searching module, a parameter collecting and processing module and a data analysis module. The walking roof searching module automatically walks along a steel rail to recognize a retarder, alignment and rail locking are completed, and position information is obtained. The parameter acquisition and processing module simulates the rolling of the retarder through a mechanical execution mechanism, and synchronously acquires and preprocesses full-work periodic energy data to obtain a multi-dimensional data set; and the data analysis module judges the fault state of the retarder based on the multi-dimensional data set, constructs a multi-factor coupling life degradation model in combination with the performance attenuation factor and the working environment factor to predict the remaining service life, generates inspection information and transmits the inspection information to the maintenance management terminal. According to the invention, automatic inspection of the retarder is realized, the detection precision and efficiency are improved, and scientific data support is provided for maintenance of the retarder.

Description

Technical Field

[0001] This invention relates to the field of railway equipment inspection technology, specifically to an automatically controlled deceleration-assisted inspection trolley. Background Technology

[0002] Railway speed reducers are core safety devices in the hump yard speed control system of railway marshalling yards, widely used for speed control during the shunting of railway vehicles. Their stability and reliability directly determine the efficiency of railway marshalling operations and traffic safety. Currently, the daily inspection and performance testing of speed reducers in domestic railway yards still mainly rely on traditional manual inspection methods. Inspectors visually inspect for visible faults such as external damage and oil leaks, and qualitatively assess the speed reducer's work capacity and rebound performance by manually stepping on it. This method not only suffers from low inspection efficiency, high labor intensity, and high operating costs, but also suffers from the significant influence of the inspectors' professional level and subjective experience on the test results. It cannot obtain quantitative test data on the speed reducer's working performance, cannot identify hidden faults such as wear and performance degradation of internal components, and cannot scientifically predict the remaining service life of the speed reducer. This makes it highly susceptible to safety accidents such as speeding during shunting and coupling conflicts due to missed faults or misjudgments, and is no longer suitable for the development needs of intelligent and unmanned operation and maintenance in modern railway yards.

[0003] To address the numerous drawbacks of manual inspection of deceleration tops, the industry has begun research and development of related automated inspection equipment. For example, Chinese patent CN214523786U discloses a PLC-based auxiliary inspection trolley for deceleration tops. This solution uses photoelectric sensors to identify the position of the deceleration top, and uses a push rod cylinder to drive a pressing rod and pressure sensor to complete the pressing detection of the deceleration top. At the same time, the mechanical transmission linkage of the pressing action is used to clamp and fix the rail, which to a certain extent realizes the semi-automation of deceleration top inspection and reduces the intensity of manual operation. However, this existing technology still has significant technical shortcomings: First, it can only acquire single-point pressure detection data through pressure sensors, and cannot simultaneously collect multi-dimensional performance data such as piston displacement, working reaction force, and return time throughout the entire working cycle of the deceleration top's downward loading, unloading, and rebound, making it difficult to accurately identify the type and severity of deceleration top faults; Second, it lacks a standardized preprocessing procedure for detection data, and the raw detection data is easily affected by the on-site environment and mechanical vibration, resulting in insufficient detection accuracy and data reliability; Third, it does not combine the deceleration top's own performance degradation law with the service environment and load conditions to build a life prediction model, making it impossible to quantitatively predict the remaining service life of the deceleration top, and making it difficult to support preventive maintenance throughout the entire life cycle of the deceleration top, thus failing to fundamentally solve the industry pain point of insufficient intelligence in the operation and maintenance of deceleration tops. Summary of the Invention

[0004] This invention provides an automatically controlled deceleration top auxiliary inspection trolley to solve the problems of existing deceleration top inspections that rely on manual handheld equipment, resulting in low work efficiency, insufficient detection accuracy and standardization, and safety risks associated with manual on-site operations, making it difficult to achieve automated and accurate monitoring of the deceleration top's performance status throughout its entire lifecycle.

[0005] To address the aforementioned problems, according to one aspect of the present invention, an automatically controlled deceleration top-assisted inspection trolley is disclosed, comprising a traveling top-finding module, a parameter acquisition and processing module, and a data analysis module; The traveling and top-finding module is used to autonomously travel along the railway rail to be measured, identify deceleration tops, align the deceleration tops, and then perform rail locking and fixation, while acquiring the position information of the deceleration tops. The parameter acquisition and processing module is used to perform simulated rolling loading operation on the deceleration jack through the mechanical actuator, and simultaneously acquire the full working cycle performance data of the deceleration jack during the simulated rolling loading operation, and store the full working cycle performance data after preprocessing. The data analysis module is used to identify the fault status of the deceleration top based on a multi-dimensional dataset, and to construct a multi-factor coupled life degradation model by combining performance degradation factor and working environment factor to predict the remaining service life of the deceleration top. Based on the remaining service life and fault status identification results, the module generates inspection information of the deceleration top and sends it to the maintenance management terminal.

[0006] Furthermore, the traveling and top-finding module includes a servo traveling mechanism, a rail locking assembly, a laser alignment unit, and a track coding and positioning unit. These components are electrically connected to form a linkage control logic. The servo traveling mechanism employs a wheeled traveling structure driven by dual servo motors, adapting to the standard gauge specifications of railway rails. It can autonomously travel along the railway rail under test based on the real-time identification signal from the laser alignment unit, and possesses the functions of constant-speed travel and millimeter-level displacement fine-tuning, providing adjustment support for the alignment of the deceleration top. The laser alignment unit is equipped with multiple sets of laser ranging sensors, performing laser ranging scans on a designated area of ​​the rail surface in a surface scanning detection mode to identify and extract the contour boundary feature points of the deceleration top mushroom head, and then using the wheels... The center position coordinates of the deceleration top are calculated using the contour fitting method. Simultaneously, the position deviation signal from the center position coordinates of the deceleration top is transmitted in real time to the control terminal of the servo traveling mechanism. This drives the servo traveling mechanism to perform position compensation and fine-tuning until the mechanical actuator of the deceleration top auxiliary inspection trolley is coaxially aligned with the mushroom head of the deceleration top, with an alignment deviation ≤1cm. After the laser alignment unit completes the alignment and outputs the alignment signal, the rail locking assembly triggers a clamping action based on the alignment signal. The clamping components clamp the railway rail from both sides towards the center, ensuring a secure rail-locked fixation for the deceleration top auxiliary inspection trolley to prevent displacement deviations during subsequent simulated rolling loading operations. The track coding and positioning unit collects the position information of the deceleration top after the rail locking assembly completes the rail-locking fixation.

[0007] Furthermore, the parameter acquisition and processing module includes a servo electric cylinder, a simulated rolling wheel, a multi-sensor acquisition unit, and a data preprocessing unit. The servo electric cylinder, simulated rolling wheel, multi-sensor acquisition unit, and data preprocessing unit form a collaborative working logic through a transmission structure or electrical connection. The mechanical actuator is linked to the electrical interlock circuit of the rail locking assembly. It can only start the simulated rolling loading operation after receiving the arrival signal output by the traveling jacking module. The simulated rolling wheel is driven vertically towards the deceleration jack mushroom head to perform the downward loading action with preset standard loading parameters (downward load, constant downward speed, preset lifting stroke, and unloading rebound rate). The unloading rebound operation is completed after the maximum lifting stroke of the deceleration jack is reached, realizing the simulation of the actual working process of the train wheel rolling the deceleration jack. The multi-sensor acquisition unit establishes a synchronous trigger signal link with the control end of the servo electric cylinder. The acquisition program is synchronously triggered at the moment the servo electric cylinder starts the downward action to collect the full working cycle performance data of the deceleration jack and transmit it to the data preprocessing unit for preprocessing.

[0008] Furthermore, the sampling frequency of the multi-sensor acquisition unit is 120Hz-360Hz, and the full working cycle performance data includes piston displacement, working reaction force, and piston return time. The preprocessing process is as follows: using the trigger signal of the servo electric cylinder starting the pressing action as a unified time reference, the piston displacement and working reaction force are first time-aligned, and then a 5th-order moving average filtering algorithm is used to perform sliding filtering noise reduction on the time-aligned piston displacement and working reaction force. Based on the Laida criterion, outliers are removed from the piston displacement and working reaction force after sliding filtering noise reduction, and linear interpolation is used simultaneously to fill in the missing values ​​after outlier removal, thus obtaining the piston displacement time series and working reaction force time series. The piston displacement time series and working reaction force time series are then fitted with a third-order polynomial curve using the least squares method to obtain a smooth and continuous displacement time series curve and reaction force time series curve, thereby obtaining a multi-dimensional dataset including piston return time, displacement time series curve, and reaction force time series curve.

[0009] Furthermore, after completing the preprocessing of the performance data for the entire working cycle and generating a multi-dimensional dataset, the data preprocessing unit synchronously outputs the multi-dimensional dataset to the local storage unit of the deceleration top auxiliary inspection trolley to complete the data solidification and storage. At the same time, after completing the validity verification of the storage of the current multi-dimensional dataset, the data preprocessing unit generates a single-top inspection completion trigger signal and transmits it to the traveling top-finding module to trigger and drive the deceleration top auxiliary inspection trolley to release the rail locking assembly and control the servo traveling mechanism to autonomously travel along the railway rail to be measured and perform the inspection measurement operation of the next deceleration top.

[0010] Furthermore, the method for determining the fault state is as follows: S1. The moment the servo electric cylinder starts its downward pressing action is taken as the zero point of time. To establish a unified time series reference for displacement and reaction time series curves, and to determine the sampling time interval. , The sampling frequency of the multi-sensor acquisition unit, and the piston return time. For the start-up and unloading time of the servo electric cylinder When the deceleration piston rebounds to its initial zero position The time interval, i.e. Piston return time in multi-dimensional datasets Corresponding timing nodes ( (Round down) as the dividing point, the displacement time series curve Segmented into non-return displacement curves ( Corresponding time sequence node ) and return displacement curve ( Corresponding time sequence node , (Total number of sampling points per cycle), reaction force time series curve Divided into non-return reaction force curves ( ) and return reaction force curve ( Based on this, four curves are obtained.

[0011] S2. By sampling time interval Discretize the four curves at equal intervals to obtain the corresponding discrete sequence, namely: the non-backward displacement discrete sequence. Discrete sequence of return displacement Non-return reaction discrete sequence Discrete sequence of return reaction force The zero-filling method is used to pad short sequences in the discrete sequence to the total number of sampling points in a single period. Consistent, each constructing dimensions as Non-return displacement matrix Return displacement matrix Non-return reaction force matrix and return reaction force matrix And concatenate four single-channel matrices vertically along the channel dimension to obtain a matrix with dimension . Multichannel fusion matrix The splicing formula is: ; S3. Based on a preset healthy deceleration top multi-dimensional dataset, the healthy deceleration top multi-dimensional data in the healthy deceleration top multi-dimensional dataset is a multi-dimensional dataset of fault-free deceleration tops, and the number of healthy deceleration top multi-dimensional data... There should be no fewer than 50 data points, and for each healthy deceleration, the maximum number of dimensions of data should be constructed using the S1-S2 method. Multi-channel standard fusion matrix of dimension, all multi-channel standard fusion matrices constitute a multi-channel standard fusion matrix set. Based on this, the mean matrix of the multi-channel standard fusion matrix set is calculated. sum and variance matrix Both are Dimension matrix; The formula for calculating the mean matrix is: ; in, The mean matrix is ​​the first Line 1 Column matrix elements, This is the matrix in the i-th row and j-th column of the multi-channel standard fusion matrix corresponding to the m-th health deceleration top multi-dimensional data; The formula for calculating the variance matrix is: ; in, The matrix element in the i-th row and j-th column of the variance matrix represents the sample variance of the corresponding matrix element of all multi-channel standard fusion matrices at the i-th channel and j-th time node, used to reflect the performance fluctuation of the healthy deceleration peak at that time position. S4. Employ the Weighted Dynamic Time Warping (WDTW) algorithm to iteratively match the multi-channel fusion matrix. With mean matrix Simultaneously calculate the multi-channel fusion matrix Channel variance Specifically, this involves initializing the weights of the four channels. ( (corresponding sequentially to non-return displacement, return displacement, non-return reaction force, and return reaction force channels), constructing... Cumulative cost matrix of dimensions ; The cumulative cost matrix matrix elements The calculation formula is: ; in, and The sequence number is the time-series node number, and the boundary conditions are: , ; The optimal regularization path is found using a dynamic time programming algorithm, yielding the total regularization cost for this iteration. Then update the weights based on the proportion of matching costs for each channel. The iterative matching process is repeated until the iteration terminates, and the termination condition is the change in weights. Or the number of iterations reaches the preset limit for the number of times the timeline is aligned with the path; The formula for calculating the channel variance is: ; in, for No. The arithmetic mean of the row sequence; S5. Channel variance based on iterative matching results and multi-channel fusion matrix variance matrix global mean Calculate the deviation of the deceleration top performance. And based on the deviation, determine the fault state of the deceleration top, when When the deceleration top is determined to be in a fault state, When the deceleration peak is determined to be in a non-fault state, the final fault state result of the deceleration peak is output; the iterative matching result is the minimum total regularization cost obtained by the iterative matching. ; The formula for calculating the deviation is: ; in, The preset maximum normalization cost threshold, For preset weighting coefficients and Preferred , , The variance matrix The arithmetic mean of all matrix elements.

[0012] Furthermore, the prediction of the remaining service life is only made when the failure state of the deceleration jack is a non-failure state; the method for predicting the remaining service life is as follows: Step a. Retrieve the historical multi-dimensional dataset of the deceleration jack using location information. Simultaneously, obtain the train operation data of the line corresponding to the deceleration jack from the railway line operation management system, and obtain the service environment data within the corresponding service cycle from the trackside environmental monitoring stations along the line. Based on the above multi-source data, extract the performance degradation factor characterizing the performance change of the deceleration jack itself, the load factor characterizing the impact of service load, and the environmental factor characterizing the impact of service environment. The working environment factor is the load factor and the environmental factor. The performance degradation factors include the piston maximum stroke attenuation rate, the working reaction force degradation rate, and the return time extension rate. The quantitative calculation formulas for each factor are as follows: Piston maximum stroke attenuation rate : In the formula, The rated maximum piston stroke of the decelerator is specified by the manufacturer. For the first The maximum piston stroke measured during the second inspection (the maximum displacement value in the displacement-time curve); the degradation rate of working reaction force. : In the formula, To reduce the rated working reaction force specified by the manufacturer, For the first Peak working reaction force measured during the second inspection (maximum value in the reaction force time-series curve); return time extension rate. : In the formula, To reduce the speed of the decelerator, the factory-specified rated piston return time is used. For the first The piston return time measured during the second inspection; The environmental factor is the orbital temperature fluctuation range. and the fluctuation range of ambient humidity The temperature fluctuation range of the rail The difference between the maximum and minimum track temperature during the statistical period, and the fluctuation range of ambient humidity. This represents the difference between the maximum and minimum relative humidity of the line environment within the statistical period. The load factor is the average daily train passing frequency of the service line. This is the ratio of the total number of trains passing through the line within the statistical period to the number of days in the statistical period.

[0013] Step b. Overall performance degradation amount composed of performance degradation factor With the degradation dependent variable and the loading factor and environmental factor as independent variables, a multivariate nonlinear degradation equation is constructed, and the baseline degradation coefficient of the model is obtained by solving the Levenberg-Marquardt multivariate nonlinear fitting algorithm. Weights of each degradation contribution , , The optimal solution is obtained, and a lifespan degradation benchmark model is constructed based on this. Its mathematical expression is: In the formula To reduce the service life of the top The corresponding overall performance degradation, The baseline degradation coefficient of the model. , , These are, respectively, the coupling terms of daily average train passing frequency and corresponding degradation contribution weight, the coupling terms of rail temperature fluctuation amplitude and corresponding degradation contribution weight, and the coupling terms of ambient humidity fluctuation amplitude and corresponding degradation contribution weight; The overall performance degradation The quantitative calculation formula is as follows: ; in, , , These are the performance weights for the piston's maximum stroke attenuation rate, working reaction force degradation rate, and return time extension rate, respectively, and they satisfy the following conditions: , , , All ≥ 0; Step c. Using the multi-dimensional dataset of the deceleration top obtained during this inspection, calculate the current real-time performance degradation factor and comprehensive performance degradation amount according to the methods in steps a and b. Input these into the lifetime degradation benchmark model constructed in step b, and use the recursive least squares method to calculate the benchmark degradation coefficient of the lifetime degradation benchmark model. Individual operating condition adaptation corrections are performed. Simultaneously, based on the real-time load factors and environmental factors of the currently serving line, the degradation contribution weights corresponding to each factor are dynamically coupled and adjusted to obtain a multi-factor coupled life degradation model adapted to the actual service state of the deceleration jack. Its mathematical expression is: In the formula This is the corrected baseline degradation factor. , , These are the real-time average daily train passing frequency of the service lines during the current statistical period, the real-time track temperature fluctuation range of the lines during the current statistical period, and the real-time environmental humidity fluctuation range of the lines during the current statistical period. , , The degradation contribution weights corresponding to each factor after dynamic coupling adjustment; Step d. Based on the multi-factor coupled life degradation model, with time as the extrapolation variable, iteratively calculate the trend of the overall performance degradation over service time until the overall performance degradation reaches the preset failure threshold. Record the corresponding total service time Total service time Compared with the current length of service The difference is taken as the remaining service life of the deceleration top. The preset failure threshold The minimum value is the critical failure value of the speed reducer specified in the industry standard TB / T 2460-2016 "Railway Speed ​​Reducers" or the performance control threshold required for the corresponding line operation safety. The final output is the calculated remaining service life, which is then bound and stored with the location information and fault status judgment results of the corresponding speed reducer and synchronously pushed to the maintenance management terminal.

[0014] Compared with the prior art, the present invention has the following advantages: This invention addresses the shortcomings of existing technologies, such as low efficiency and subjective results of manual inspections, and the limitations of automated inspection equipment in terms of single data acquisition, lack of standardized preprocessing, and inability to quantify and predict remaining service life. Through an integrated modular design, it achieves precise alignment and locking of the deceleration top and standardized acquisition and preprocessing of multi-dimensional performance data. This automatically triggers continuous inspection operations and, based on algorithms, completes fault quantification and multi-factor coupled remaining service life prediction. This realizes automation, standardization, and quantification of inspections, significantly improving inspection efficiency and detection accuracy, and providing reliable support for intelligent preventative maintenance of railway deceleration tops. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a structural diagram of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] In one embodiment, such as Figure 1 The aforementioned automatic control deceleration top-assisted inspection trolley includes a traveling top-finding module, a parameter acquisition and processing module, and a data analysis module; The traveling and top-finding module is used to autonomously travel along the railway rail to be measured, identify deceleration tops, align the deceleration tops, and then perform rail locking and fixation, while acquiring the position information of the deceleration tops. The parameter acquisition and processing module is used to perform simulated rolling loading operation on the deceleration jack through the mechanical actuator, and simultaneously acquire the full working cycle performance data of the deceleration jack during the simulated rolling loading operation, and store the full working cycle performance data after preprocessing. The data analysis module is used to identify the fault status of the deceleration top based on a multi-dimensional dataset, and to construct a multi-factor coupled life degradation model by combining performance degradation factor and working environment factor to predict the remaining service life of the deceleration top. Based on the remaining service life and fault status identification results, the module generates inspection information of the deceleration top and sends it to the maintenance management terminal.

[0019] Furthermore, the following example, using the inspection operation of the 60kg / m rail-mounted deceleration top on a hump yard speed regulation line at a railway marshalling yard, fully illustrates the working process of the automatically controlled deceleration top auxiliary inspection trolley of the present invention: The vehicle's traveling and jacking module, parameter acquisition and processing module, and data analysis module are integrated into the vehicle body and are linked and controlled by the on-board main control unit. The vehicle body is adapted to the standard gauge of railway rails and is powered by a lithium battery pack to meet the needs of continuous 8-hour outdoor inspection operations. The modules work together to complete the entire process of autonomous positioning of the deceleration jack, rail locking and fixing, simulated rolling, data acquisition and preprocessing, fault diagnosis, life prediction and inspection information output.

[0020] The traveling and top-finding module includes a servo traveling mechanism, a rail locking assembly, a laser alignment unit, and a track coding and positioning unit. These components are electrically connected to form a linked control logic. The servo traveling mechanism employs a wheeled traveling structure driven by dual servo motors with a rated motor speed of 300 r / min, suitable for standard gauge 60 kg / m rails. The onboard main control unit pre-imports the inspection path for the hump track of the railway marshalling yard. The servo traveling mechanism drives the inspection trolley to autonomously travel along the rail to be tested at a speed of 0.5 m / s. The laser alignment unit is equipped with four sets of laser ranging sensors, with a detection range of 0-50 cm and an accuracy of ±0.1 mm. It performs laser ranging scans on a designated 30 cm × 30 cm area of ​​the rail surface using a surface scanning detection mode. When the deceleration top mushroom head is scanned, its contour boundary feature points are extracted, and the coordinates of the center position of the deceleration top are calculated using the least squares contour fitting method. The position deviation signal is transmitted in real time to the servo traveling mechanism control terminal, driving the servo traveling mechanism to perform millimeter-level displacement fine-tuning until the mechanical actuator of the inspection trolley and the mushroom head of the deceleration top are coaxially aligned, with the alignment deviation controlled within 0.8cm. After the laser alignment unit outputs the alignment signal, the rail locking assembly immediately starts the clamping action. Its electric clamping parts apply a clamping force of 80kN from both sides of the rail towards the center, firmly fixing the inspection trolley to the rail, avoiding displacement deviation caused by subsequent simulated rolling loading operations. After the rail locking is completed, the alignment signal is output to the parameter acquisition and processing module. At the same time, the track coding and positioning unit collects the position information of the deceleration top (3 tracks of the lower peak line of the hump, rail mileage K1+256.32m, 0.3m to the left of the trackside), encodes the position information, and transmits it to the on-board main control unit for storage, providing a basis for subsequent inspection information labeling.

[0021] The parameter acquisition and processing module includes a servo electric cylinder, a simulated rolling wheel, a multi-sensor acquisition unit, and a data preprocessing unit. These components form a collaborative working logic through transmission structures or electrical connections. The servo electric cylinder and the simulated rolling wheel constitute a mechanical actuator, which is electrically interlocked with the rail locking assembly, activating only upon receiving a rail locking signal. In this embodiment, the servo electric cylinder drives the simulated rolling wheel vertically towards the deceleration head of the deceleration unit using preset standard loading parameters (downward load 20kN, constant downward speed 50mm / s, preset lifting stroke 40mm, unloading and rebound rate 80mm / s). After the simulated rolling wheel completes the downward loading to its maximum lifting stroke, it performs an unloading and rebound operation according to a preset pattern, simulating the entire process of a train wheel rolling over the deceleration head. The sensor acquisition unit establishes a synchronous trigger signal link with the servo electric cylinder control terminal. The acquisition program is synchronously triggered the instant the servo electric cylinder starts the pressing action. In this embodiment, the sampling frequency of the multi-sensor acquisition unit is set to 240Hz, and the sampling time interval Δt=1 / 240≈0.00417s. This unit integrates displacement sensor, pressure sensor and time sensor, and synchronously acquires the performance data of the deceleration top from the initial static state, pressing loading, unloading and rebound to the completion of the reset of the entire working cycle. Specifically, it includes piston displacement, working reaction force and piston return time. The core measured values ​​obtained from a single simulated rolling operation are the maximum piston stroke of 38.2mm, peak working reaction force of 19.5kN and piston return time of 0.85s. After the acquisition is completed, the raw data is transmitted to the data preprocessing unit through the high-speed CAN bus.

[0022] The data preprocessing unit receives the full-cycle performance data transmitted by the multi-sensor acquisition unit and performs preprocessing: using the trigger signal of the servo electric cylinder initiating the downward pressing action as the unified time zero point. The piston displacement and working reaction force signals were sequentially aligned in time, denoised by sliding filter, removed outliers, filled with missing values, and fitted with curves to obtain a smooth and continuous displacement-time curve. and reaction force time series curve By combining the measured piston return time, a multi-dimensional dataset is formed, including piston return time, displacement time series curve, and reaction force time series curve. The data preprocessing unit synchronously outputs the generated multi-dimensional dataset to the local storage unit (SD card + solid-state drive) of the inspection trolley to complete the data solidification and storage and perform validity verification. After the verification is completed, a single-top inspection completion trigger signal is generated and transmitted to the traveling top-finding module to drive the rail locking assembly to release the rail locking fixation. At the same time, it controls the servo traveling mechanism to continue to travel autonomously along the railway rail to be measured and perform the inspection measurement operation of the next deceleration top, realizing continuous automated inspection of the deceleration top.

[0023] The data analysis module retrieves a multi-dimensional dataset from the local storage unit and performs fault state determination of the deceleration top according to a preset method: the moment when the servo electric cylinder starts its downward pressing action is taken as the zero point of time. Determine the displacement time series curve and reaction force time series curve A unified timing reference is used, with a sampling time interval Δt≈0.00417s, defining the piston return time. For the start-up and unloading time of the servo electric cylinder (t=0.9s) until the deceleration top piston rebounds to its initial zero position The time interval (t=1.75s) was calculated to yield... Corresponding time sequence node ⌊ / Δt⌋=203, with The displacement time series curve is used as the dividing point. Segmented into non-return displacement curves (Corresponding to timing nodes 1~203) and return displacement curve (Corresponding to time series nodes 204~420, total number of sampling points per cycle) ), and simultaneously the reaction time series curve Divided into non-return reaction force curves and return reaction force curve Four independent time-series curves for the working stages were obtained; the four curves were discretized at equal intervals according to the sampling time interval Δt to obtain the non-backflow displacement discrete sequence. Discrete sequence of return displacement Non-return reaction force discrete sequence Discrete sequence of return reaction force The short sequence is padded with zeros to match the total number of sampling points in a single period. Consistent, each constructing dimensions as Non-return displacement matrix Return displacement matrix Non-return reaction force matrix Return reaction force matrix By vertically concatenating four single-channel matrices along the channel dimension, we obtain a matrix with dimension [missing information]. Multichannel fusion matrix The splicing formula is: In this embodiment, the preset healthy deceleration head multi-dimensional dataset consists of 60 healthy deceleration head multi-dimensional datasets of the same model without faults. The multi-dimensional dataset for each healthy deceleration head is constructed using the method described above. A multi-channel standard fusion matrix of dimensions is formed to create a multi-channel standard fusion matrix set. The mean matrix is ​​calculated based on this matrix set. sum and variance matrix (All are) (Dimension), using the Weighted Dynamic Time Warping (WDTW) algorithm to iteratively match the multi-channel fusion matrix. With mean matrix Initialize the weights of the four channels. , build Cumulative cost matrix of dimensions The total regularization cost of this matching is obtained by solving for the optimal regularization path through dynamic time programming. Then update the weights based on the proportion of matching costs for each channel. The matching process is repeated iteratively. In this embodiment, the weight change occurs at the 8th iteration. If the iteration termination condition is met, the minimum total regularization cost is extracted. Simultaneous calculation of multi-channel fusion matrix Channel variance In this embodiment, the calculation is as follows: Calculate the variance matrix global mean Based on minimum total regularization cost Channel variance Global mean Calculate the performance deviation of the deceleration top ≈0.449<0.58, therefore the deceleration top is determined to be in a non-faulty state; After determining the fault status, the data analysis module constructs a multi-factor coupled life degradation model to predict the remaining service life of the deceleration top by combining its performance degradation factor and operating environment factors. This model retrieves the historical multi-dimensional dataset of the deceleration top over the past three years using location information collected by the track coding and positioning unit. Simultaneously, it obtains train operation data for the corresponding service line from the railway line operation management system and environmental data for the corresponding three-year service cycle from trackside environmental monitoring stations. Based on this multi-source data, performance degradation factors, load factors, and environmental factors are extracted. The performance degradation factor includes the piston's maximum stroke attenuation rate. Work reaction degradation rate Return trip time extension rate The factory-calibrated rated value of this deceleration top is , , The measured value during this inspection was , , Calculated , , Environmental factors include orbital temperature fluctuation amplitude. and the fluctuation range of ambient humidity The statistical period is the past 3 months, and the calculation is as follows: =45℃ (maximum rail temperature 60℃, minimum rail temperature 15℃) (Maximum humidity 85%RH, minimum 25%RH), load factor is the average daily train passing frequency of the service line. The statistical period is the past 3 months, with a total of 2790 trains passing through the line and a statistical period of 90 days. Calculations are as follows: Columns / day; Overall performance degradation amount composed of performance degradation factor The dependent variable is degenerate, and the calculation formula is: The performance weight is taken as follows: , , ( Substituting the data, we can calculate... ≈7.09%, with loading factor Environmental factors , Using historical multi-source data as the independent variable, the Levenberg-Marquardt multivariate nonlinear fitting algorithm was used to solve for the degradation contribution weights of each factor, and a baseline model for lifespan degradation was constructed. In this embodiment, the model parameters obtained by fitting are the baseline degradation coefficients. Loading factor weights Weight of rail temperature fluctuation amplitude Humidity fluctuation range weight The real-time performance degradation factor, overall performance degradation, and current real-time load factor obtained during this inspection will be used as data. Series / day, environmental factors =45℃ Input the lifespan degradation baseline model, and use the recursive least squares method to calculate the baseline degradation coefficients. Perform individual working condition adaptation corrections, and obtain the corrected result. Simultaneously, based on the current real-time operating conditions, the degradation contribution weights of each factor are dynamically coupled and adjusted, resulting in... , , Finally, a multi-factor coupled life degradation model adapted to the actual service state of the deceleration top was obtained. In this embodiment, a preset failure threshold is used. Taking the minimum value as 25% of the performance control threshold for the operational safety requirements of this railway marshalling yard, and using time as the extrapolation variable, Substituting into the multi-factor coupled lifetime degradation model, the total service time when the comprehensive performance degradation reaches the failure threshold is calculated iteratively. The deceleration device has been in service for [number] years. years, therefore remaining service life Year.

[0024] The data analysis module binds the location information of the deceleration top (3 tracks of the lower peak line, K1+256.32m), the fault status judgment result (non-fault status), and the remaining service life (5.5 years) to generate standardized inspection information. This inspection information is stored in the local storage unit of the trolley and transmitted in real time to the maintenance management terminal of the railway station through its 5G wireless communication module. The maintenance management terminal visualizes the inspection information and classifies and labels the deceleration tops according to fault status and remaining service life, providing scientific data support for the operation and maintenance management of deceleration tops in the railway station: deceleration tops in fault status are marked with a red warning and prompt immediate replacement; deceleration tops with less than 1 year of remaining service life are marked with a yellow warning and prompt planned maintenance; and deceleration tops that are not faulty and have sufficient remaining service life are marked with green normal and inspected according to the regular cycle.

[0025] The above description is merely a specific embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. An automatically controlled deceleration-assisted inspection trolley, characterized in that, include: The traveling and top-finding module is used to autonomously travel along the railway rail to be measured, identify deceleration tops, align the deceleration tops, and then perform rail locking and fixation, while acquiring the position information of the deceleration tops. The parameter acquisition and processing module is used to perform simulated rolling loading operation on the deceleration jack through the mechanical actuator, and simultaneously acquire the full working cycle performance data of the deceleration jack during the simulated rolling loading operation, and store the full working cycle performance data after preprocessing. The data analysis module is used to identify the fault status of the deceleration top based on a multi-dimensional dataset, and to construct a multi-factor coupled life degradation model by combining performance degradation factor and working environment factor to predict the remaining service life of the deceleration top. Based on the remaining service life and fault status identification results, the module generates inspection information of the deceleration top and sends it to the maintenance management terminal.

2. The automatically controlled deceleration-assisted inspection trolley according to claim 1, characterized in that, The traveling and top-finding module includes a servo traveling mechanism, a rail locking assembly, a laser alignment unit, and a track positioning unit. The servo traveling mechanism drives the deceleration top auxiliary inspection trolley to travel autonomously along the railway rail to be tested. The laser alignment unit uses laser ranging to scan the mushroom-shaped outline boundary of the deceleration top to determine the center position of the deceleration top and drives the servo traveling mechanism to complete the alignment of the mechanical actuator of the deceleration top auxiliary inspection trolley with the deceleration top. The rail locking assembly clamps the railway rail after alignment to fix the rail locking of the deceleration top auxiliary inspection trolley. The track positioning unit collects the position information of the deceleration top after alignment.

3. The automatically controlled deceleration-assisted inspection trolley according to claim 2, characterized in that, The alignment deviation is ≤1cm. The rail locking assembly and the parameter acquisition and processing module are electrically interlocked. The parameter acquisition and processing module can only start the simulated rolling loading operation after the rail locking assembly completes the rail locking and outputs the positioning signal.

4. The automatically controlled deceleration-assisted inspection trolley according to claim 1, characterized in that, The parameter acquisition and processing module includes a servo electric cylinder, a simulated rolling wheel, a multi-sensor acquisition unit, and a data preprocessing unit. The servo electric cylinder and the simulated rolling wheel are connected to form the mechanical actuator, which is used to receive a position signal and drive the simulated rolling wheel to perform simulated rolling operations of pressing down and unloading and rebounding on the deceleration top with preset standard loading parameters. The multi-sensor acquisition unit is synchronously triggered with the servo electric cylinder to acquire the full working cycle performance data of the deceleration top, and transmits the full working cycle performance data to the data preprocessing unit for preprocessing.

5. The automatically controlled deceleration top-assisted inspection trolley according to claim 4, characterized in that, The sampling frequency of the multi-sensor acquisition unit is 120Hz-360Hz. The full working cycle performance data includes piston displacement, working reaction force, and piston return time. The preprocessing process is as follows: the piston displacement and working reaction force are sequentially aligned, denoised by sliding filter, removed outliers, filled with missing values, and fitted with curves to obtain displacement time-series curves and reaction force time-series curves. Based on this, a multi-dimensional dataset including piston return time, displacement time-series curves, and reaction force time-series curves is obtained.

6. The automatically controlled deceleration-assisted inspection trolley according to claim 5, characterized in that, The method for determining the fault status is as follows: S1. Using the piston return time in the multidimensional dataset as the dividing point, the displacement time series curve is divided into the non-return displacement curve and the return displacement curve, and the reaction force time series curve is divided into the non-return reaction force curve and the return reaction force curve, resulting in four curves. S2. Discretize the four curves at equal intervals according to the sampling frequency, construct the non-back displacement matrix, back displacement matrix, non-back reaction force matrix and back reaction force matrix, and splice them to obtain the multi-channel fusion matrix; S3. Based on the preset multi-dimensional dataset of health deceleration items, construct a multi-channel standard fusion matrix set for the multi-dimensional dataset of health deceleration items according to S1-S2, and calculate the mean matrix and variance matrix of the multi-channel standard fusion matrix set accordingly. S4. The weighted dynamic time warping algorithm is used to iteratively match the multi-channel fusion matrix and the mean matrix, and the channel variance of the multi-channel fusion matrix is ​​calculated at the same time. S5. Calculate the deviation based on the iterative matching results, channel variance, and variance matrix, and determine the fault state of the deceleration top according to the deviation, thus obtaining the fault state result of the deceleration top.

7. The automatically controlled deceleration-assisted inspection trolley according to claim 6, characterized in that, The iteration termination condition for the iterative matching is that the weight change is less than 0.01 or the number of iterations reaches a preset upper limit. If the deviation is ≥0.58, the deceleration cap is determined to be in a fault state; otherwise, the deceleration cap is determined to be in a non-fault state.

8. The automatically controlled deceleration-assisted inspection trolley according to claim 1, characterized in that, The method for predicting the remaining useful life is as follows: Step a. Obtain the historical multi-dimensional dataset of the deceleration jack and the train operation data and service environment data of the corresponding service line of the deceleration jack. Extract the performance degradation factor that characterizes the performance change of the deceleration jack itself, the load factor that characterizes the impact of service load, and the environmental factor that characterizes the impact of service environment. The working environment factor is the load factor and the environmental factor. Step b. Using the performance degradation factor as the degradation dependent variable and the load factor and environmental factor as independent variables, solve the degradation contribution weights of each factor through a multivariate nonlinear fitting algorithm to construct a life degradation benchmark model. Step c. Input the current multi-dimensional dataset of the deceleration pylon into the service life degradation benchmark model, and perform individual working condition adaptation correction on the degradation coefficient of the service life degradation benchmark model to obtain a multi-factor coupled service life degradation model that adapts to the actual service state of the deceleration pylon. Step d. Based on the multi-factor coupled life degradation model, calculate the remaining service time when the performance degradation of the deceleration top reaches the preset failure threshold, and output the remaining service life of the deceleration top.

9. The automatically controlled deceleration top-assisted inspection trolley according to claim 8, characterized in that, The performance degradation factor includes the piston maximum stroke attenuation rate, working reaction force degradation rate, and return time extension rate; the environmental factor is the rail temperature fluctuation range and the ambient humidity fluctuation range; and the load factor is the average daily train passing frequency of the service line.

Images