Performance testing and evaluation method for door motors of urban rail vehicles

Through the support vector data description model, the performance characteristics parameters of the door motor of urban rail vehicles were analyzed, which solved the shortcomings of early fault detection and performance evaluation, and realized quantitative evaluation and status monitoring of door motor performance, providing effective maintenance decision support. Improve vehicle operation safety and service quality.

CN114021452BActive Publication Date: 2024-11-19CRRC CHANGCHUN RAILWAY VEHICLES CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111288095.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-02
Publication Date
2024-11-19
Estimated Expiration
2041-11-02

AI Technical Summary

Technical Problem

The existing technology lacks effective methods in early fault detection and overall performance evaluation of door motors of urban rail vehicles, resulting in insufficient support for fault prediction and maintenance decisions, and it is difficult to avoid parking and safety problems caused by motor failure.

Method used

Using the Support Vector Data Description (SVDD) model, the performance characteristic parameters and reference values ​​of the door motor are defined, the deviation between the performance indicators and reference values ​​are calculated, and the performance indicators and performance evaluations are carried out, and the hyperspherical radius is constructed using Gaussian kernel functions and optimization problems. , realize quantitative evaluation and status monitoring of door motor performance.

Benefits of technology

It realizes detection of the performance degradation of the door motor of urban rail vehicles and early failures, provides strong maintenance decision support, reduces parking and safety risks caused by failures, and does not require a large number of fault samples for data analysis, improving the ability of the method Operative.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114021452B_ABST
    Figure CN114021452B_ABST
Patent Text Reader

Abstract

A method for detecting and evaluating the performance of a door motor of an urban rail vehicle includes: 1) defining the current peak value of the door motor, namely the current peak value of the acceleration section, the current peak value of the high-speed section and the current peak value of the deceleration section, as the characteristic parameters of the door motor performance; 2) constructing the door motor performance index using support vector data description (SVDD); 3) defining the SVDD hypersphere radius obtained by sample training using the motor performance characteristic parameters when the door motor is fault-free, has no performance degradation and the door is opened and closed normally as the reference value of the door motor performance index; 4) calculating the deviation between the door performance index value and the reference value, and 5) performing performance detection and performance evaluation based on the deviation value. By analyzing the deviation between the performance index of the door motor of an urban rail vehicle and its reference value during actual operation, the abnormal operation of the system is detected, and then the early fault is identified; at the same time, the performance status of the door motor can be quantitatively evaluated to provide support for the realization of state-based maintenance decisions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for detecting and evaluating the performance of a door motor of an urban rail vehicle based on support vector data description and fuzzy judgment, and belongs to the technical field of health management of urban rail vehicles. Background Art

[0002] The door system is a key subsystem of urban rail vehicles. Its reliability and safety are directly related to the safety of passengers and the quality of driving services. Therefore, it is a key issue in the operation and management of urban rail vehicles. The drive motor is a key component that provides driving force for the door switch. Its performance status determines the operating performance and safety of the door. Therefore, it is necessary to monitor the operating status of the door drive motor, detect the fault of the door motor in time, and evaluate its performance status to provide decision support for the maintenance and replacement of the motor, so that it can be processed before the motor fails, avoiding accidents such as parking and clearing passengers caused by door failure caused by motor failure, improving the quality of driving services, and ensuring driving safety.

[0003] At present, the door systems of urban rail vehicles are equipped with a large number of sensors to monitor their working parameters, and have certain diagnostic capabilities, which can detect some faults. In addition, the daily maintenance of urban rail vehicles can also actively discover and prevent various faults, including those in the door system. However, the current status detection and daily maintenance of urban rail vehicle door motors focus on the diagnosis and repair of faults that have already occurred, and there is still a lack of applicable methods and technologies for early detection of faults. At the same time, there is also a lack of feasible methods for evaluating the overall performance of door motors and analyzing and evaluating the system performance degradation process. Summary of the invention

[0004] The purpose of the present invention is to provide a method for detecting the performance of a door motor of an urban rail vehicle. By analyzing the deviation of the performance index of the door motor of an urban rail vehicle during actual operation from its reference value, the abnormal operation of the system is detected, and then the early failure of the system is identified; at the same time, the performance state of the door motor can be quantitatively evaluated based on the deviation of the actual performance index from the expected value, so as to provide support for the realization of maintenance decision-making based on the state.

[0005] To achieve the above-mentioned object of the invention, the present invention provides a method for detecting and evaluating the performance of a door motor of an urban rail vehicle, which is characterized by comprising the following steps:

[0006] 1) Define the current peak value of the door motor in the acceleration section, the current peak value of the high speed section and the current peak value of the deceleration section as the performance characteristic parameters of the door motor;

[0007] 2) Use support vector data description (SVDD) to construct the performance index of the door motor;

[0008] 3) Define the SVDD hypersphere radius obtained by sample training using motor performance characteristic parameters when the door motor is fault-free, has no performance degradation, and the door opens and closes normally as the door motor performance index benchmark value;

[0009] 4) Calculate the deviation between the door performance index value and the reference value;

[0010] 5) Perform performance testing and performance evaluation based on the deviation value.

[0011] Preferably, step 3) constructs the SVDD hypersphere radius obtained by sample training by constructing the following optimization problem:

[0012]

[0013] st||X i -a|| 2 ≤R 2 +ε i ;

[0014] ε i ≥0

[0015] The dual problem of the above optimization problem is obtained:

[0016]

[0017] st0≤α i ≤C;

[0018]

[0019] Wherein, K(.) is a kernel function, and the present invention adopts a Gaussian kernel function; training samples are used for training to obtain the center and radius of the hypersphere:

[0020]

[0021] Furthermore, based on the SVDD model trained in step 3), a new sample X is constructed for the door motor characteristic parameters defined in step 1). new , calculate the distance d between the new sample and the center of the SVDD hypersphere new ,

[0022]

[0023] Calculate d new and benchmark performance indicators Deviation value ΔR:

[0024]

[0025] Preferably, the performance of the door drive motor is tested based on ΔR and the time series formed by it, and the specific method is as follows:

[0026] 1) If ΔR exceeds a given threshold δ, it indicates that the performance of the door drive motor of the urban rail vehicle exceeds the limit and needs to be repaired to eliminate faults or other factors;

[0027] 2) If the ΔR time series changes suddenly, it indicates that the door drive motor of the urban rail vehicle may have a fault and needs to be repaired and troubleshooted;

[0028] 3) If the ΔR time series shows a trend, it indicates that the door drive motor of the urban rail vehicle has failed or its performance is steadily declining, which requires close attention and fault diagnosis.

[0029] Preferably, the health quantitative assessment of the urban rail vehicle braking system is performed according to the size of ΔR. The specific method is as follows: based on the ΔR calculated in step 4), the performance of the door drive motor is evaluated; since each door opening and closing can form a sample point, the performance evaluation can be performed each time the door is opened and closed; considering that the performance degradation of the door drive motor is a slow-changing process, the performance evaluation can be performed in a longer period, such as once a day, and ΔR is first normalized, as follows:

[0030]

[0031] Among them, δ is the normalization threshold, α is the normalization factor, which is between 0 and 1, and k can be selected according to the actual situation; then the membership function is selected, and the performance of the door drive motor is evaluated according to the normalized ΔR in step 4).

[0032] Preferably, the performance evaluation of the door drive motor adopts a triangular membership function, the normalized ΔR is calculated to calculate the membership of the predefined motor grades, and the grade with the largest membership is taken as the performance grade of the door drive motor.

[0033] Compared with the existing urban rail vehicle door motor fault detection and performance evaluation technology, the present invention can make full use of the actual operation data analysis technology to detect the performance degradation and early faults of the urban rail vehicle door motor, thereby providing strong support for the condition maintenance of the urban rail vehicle braking system. In addition, in the data analysis modeling process, only the normal operation data of the urban rail vehicle door motor is needed to construct the training samples, without a large number of fault samples, making the data easier to obtain, and it is not necessary to obtain the model training data through fault injection tests, etc., which increases the operability of the method of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the working phases of the door motor;

[0035] Figure 2 It is a performance testing and evaluation flow chart. DETAILED DESCRIPTION

[0036] Reference Figure 1 and Figure 2 The specific implementation process of the present invention is as follows:

[0037] 1) According to the working characteristics of the door system and drive motor of urban rail vehicles, the door drive motor current during the door opening and closing process is divided into three data segments: acceleration segment current, high-speed segment current and deceleration segment current. The current peak values ​​in the three segments are taken as characteristic parameters, namely, the acceleration segment current peak value, the high-speed segment current peak value and the deceleration segment current peak value, recorded as I pa , I ph , I ps ;

[0038] 2) Define the performance index of the door drive motor of the urban rail vehicle, and use the support vector data description (SVDD) model to construct the performance index. The model is based on the characteristic parameters I defined in step 1). pa , I ph , I ps The sample points X i = {I pa , I ph , I ps} is used as input, and the distance d between the sample point and the center a of the SVDD hypersphere is used as the performance indicator;

[0039] 3) Define the door drive motor as normal behavior when there is no fault or performance degradation and the door is opened normally. The sample set {X i} n Train the SVDD model defined in step 2) and use the SVDD hypersphere radius as the benchmark performance indicator for the normal operation of the door drive motor Where n is the number of samples consisting of characteristic parameters that meet the definition of normal behavior.

[0040] Specifically, the following optimization problem is constructed:

[0041]

[0042] st||X i -a|| 2 ≤R 2 +ε i ;

[0043] ε i ≥0

[0044] The dual problem of the above optimization problem is obtained:

[0045]

[0046] st0≤α i ≤C;

[0047]

[0048] Wherein, K(.) is a kernel function, and the present invention adopts a Gaussian kernel function. The training samples are used for training to obtain the center and radius of the hypersphere:

[0049]

[0050] From the above formula, we can see that each support vector X α Each corresponds to a hypersphere center radius, so the present invention uses a 96% statistical control limit as the final hypersphere radius and the benchmark performance index Statistical control limits can be adjusted according to actual application.

[0051] 4) Based on the SVDD model trained in step 3), a new sample X is constructed for the door motor characteristic parameters defined in step 1) new , calculate the distance d between the new sample and the center of the SVDD hypersphere new ,

[0052]

[0053] Calculate d new and benchmark performance indicators Deviation value ΔR:

[0054]

[0055] 5) Based on ΔR and its time series, the performance of the door drive motor is tested. The performance test process is as follows: Figure 2 shown.

[0056] The specific method is: analyze the time series of ΔR and its components:

[0057] 5.1) If ΔR exceeds a given threshold δ, it indicates that the performance of the door drive motor of the urban rail vehicle exceeds the limit and needs to be repaired to eliminate faults or other factors;

[0058] 5.2) If the ΔR time series changes suddenly, it indicates that the door drive motor of the urban rail vehicle may have a fault and needs to be repaired and troubleshooted;

[0059] 5.3) If the ΔR time series shows a trend, it indicates that the door drive motor of the urban rail vehicle has failed or its performance is steadily declining, which requires close attention and fault diagnosis.

[0060] In addition, the health quantitative assessment of the braking system of urban rail vehicles can be carried out according to the size of ΔR, providing decision support for the maintenance of urban rail vehicles.

[0061] 6) Based on the ΔR calculated in step 4), the performance of the door drive motor is evaluated. Since each time the door is opened and closed, a sample point is formed, the performance evaluation can be performed each time the door is opened and closed. Considering that the performance degradation of the door drive motor is a slow process, the performance evaluation can be performed in a longer period, such as once a day.

[0062] The specific method is: first normalize ΔR, as follows:

[0063]

[0064] Among them, δ is the normalization threshold, α is the normalization factor, which is between 0 and 1, and k can be specifically selected according to the actual situation. Then, a membership function is selected, and the performance of the door drive motor is evaluated according to the normalized ΔR in step 4). The present invention adopts a triangular membership function (in actual application, other membership functions can be selected according to the performance degradation of the door drive motor), calculates the normalized ΔR, calculates the membership of the pre-defined motor level, and takes the level with the largest membership as the performance level of the door drive motor.

Claims

1. A method for detecting and evaluating the performance of door motors of urban rail vehicles. Features The following steps are involved: 1) Define the current peak value of the door motor in the acceleration section, the current peak value of the high speed section and the current peak value of the deceleration section as the performance characteristic parameters of the door motor; 2) Use support vector data description (SVDD) to construct the performance index of the door motor; 3) Define the SVDD hypersphere radius obtained by sample training using the door motor performance characteristic parameters when the door motor is fault-free, has no performance degradation, and the door opens and closes normally as the door motor performance index benchmark value; 4) Calculate the deviation between the door motor performance index value and the reference value; 5) Perform performance testing and performance evaluation based on the deviation value; Step 3) The trained SVDD model constructs a new sample Xnew using the door motor performance characteristic parameters defined in step 1), and calculates the distance dnew between the new sample and the center of the SVDD hypersphere. Calculate the deviation value ΔR between dnew and the reference value of the door motor performance index; The performance of the door motor is tested based on ΔR and its time series. The specific method is as follows: (1) If ΔR exceeds a given threshold δ, it indicates that the performance of the urban rail vehicle door motor exceeds the limit and needs to be repaired to eliminate faults or other factors; (2) If the ΔR time series changes suddenly, it indicates that the door motor of the urban rail vehicle has a fault and needs to be repaired and troubleshooted; (3) If the ΔR time series shows a trend, it indicates that the door motor of the urban rail vehicle has a fault or its performance is steadily declining, and it is necessary to pay close attention and perform fault diagnosis; The performance of the door motor is evaluated based on ΔR. The specific method is as follows: since each time the door is opened and closed forms a sample point, a performance evaluation is performed each time the door is opened and closed; considering that the performance degradation of the door motor is a slow-changing process, the performance evaluation is performed once a day. First, ΔR is normalized as follows: Wherein, δ is a normalization threshold, α is a normalization factor, which is between 0 and 1, and k is selected according to the actual situation; then the membership function is selected, and the door motor performance is evaluated according to the normalized ΔR in step 4); The door motor performance evaluation adopts a triangular membership function, calculates the normalized ΔR, calculates the membership degree of the predefined motor level, and takes the level with the largest membership degree as the performance level of the door motor.