DPC-based method and system for detecting abnormality of multi-door system of rail vehicle

An anomaly detection and rail vehicle technology, applied in the direction of railway vehicle testing, etc., can solve the problems of different parts, many parts, and inability to detect the abnormality of subway vehicle door system, and achieve good universality and reduce repetition.

Inactive Publication Date: 2019-02-22
NANJING KANGNI MECHANICAL & ELECTRICAL +1
3 Cites 5 Cited by

AI-Extracted Technical Summary

Problems solved by technology

The existing performance degradation prediction methods of other systems, such as the research on the performance degradation of fuel cells, mainly measure the AC impedance and compare the measured AC impedance value with the degradation reference value to evaluat...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Method used

[0081] After the present embodiment, the statistical assignment results are sorted out, and compared with historical fault information, the validity of the algorithm is verified. The present invention adopts a data-based mathematical modeling method, performs mathematical modeling with a density peak clustering algorithm, and e...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Abstract

The invention discloses a DPC-based method and system for detecting abnormality of a multi-door system of a rail vehicle. The method comprises the following steps that: feature extraction is carried out on main track data of a vehicle door system, each door opening-closing process is segmented, time-domain feature extraction and frequency-domain feature extraction are carried out respectively on aplurality of motor parameter value of each segment, the time-domain features and frequency-domain features are combined to generate system state variables of the vehicle door system; on the basis ofa DPC method, a system multi-door health degree model is established for the extracted system state variables; and an e abnormal state of the vehicle door is identified by using an Euclidean distance.According to the invention, mathematical modeling is carried out based on a density peak clustering algorithm and the health degree model of the multi-vehicle door system is established; periodical horizontal comparison is carried out on the obtained multi-vehicle door model by using the established model to complete abnormality detection of the vehicle multi-door system. Therefore, the repetitive experimental design and data collection work are reduced. The density peak clustering algorithm is applied to the rail vehicle door fault detection technology first time; the influence on the vehicle door type or locking device type is eliminated; and the universality is high.

Application Domain

Technology Topic

Image

  • DPC-based method and system for detecting abnormality of multi-door system of rail vehicle
  • DPC-based method and system for detecting abnormality of multi-door system of rail vehicle
  • DPC-based method and system for detecting abnormality of multi-door system of rail vehicle

Examples

  • Experimental program(1)

Example Embodiment

[0043] The present invention will be further described below in conjunction with the drawings. The following embodiments are only used to illustrate the technical solutions of the present invention more clearly, and cannot be used to limit the protection scope of the present invention.
[0044] figure 1 It is a flowchart of an abnormality monitoring method for a vehicle multi-door monitoring system according to a specific embodiment of the present invention; refer to figure 1 It shows that the present embodiment performs health modeling based on the cumulative data of the positive line on multiple doors of a train to obtain a health model of the multi-door system, and then performs periodic horizontal comparison between the doors to mark the abnormal doors, Complete the automatic identification of the door health status. The whole process includes the following steps:
[0045] Step (A), collecting and preprocessing the main line data of multiple door systems on a subway train;
[0046] It should be noted that the data in this embodiment is the actual operating data of N doors over a period of time, where the positive line data is the data collected by the system motor sensor; in other specific embodiments, the data source can also be uploaded by rail vehicles To the data stored in the system; in addition, if the data to be processed has been processed, the method of the present invention can directly perform the following operations.
[0047] The data preprocessing includes alignment of the positive line data and removal of problem data. The problem data includes data less than a preset normal data threshold and data outside the normal data range.
[0048] The data collected in this embodiment are the three process variables of motor rotation angle, speed, and current. Therefore, the data preprocessing is to correct the data collected by the vehicle door system, to align the normal data and remove the problem data. The number of sampling points including speed, rotation angle and current data is significantly less than the normal sampling number, or the data whose initial rotation angle is not within the normal range. Time domain feature extraction. Extract 6 time domain features for each segment, including extracting each segment of the motor separately The maximum, minimum, mean, variance, skewness, and kurtosis of the parameter values ​​constitute the time domain feature set of the system. For example, the average value of the rising speed, high speed, slowing speed, slow-moving section and post-positioning section of the door opening and closing process is calculated ( What is obtained is 2×3×5=30 characteristic variables. Similarly, the maximum value, minimum value, mean value, variance, skewness and kurtosis are 180 in total, using a two-dimensional matrix X∈R n×p To express that X=X 1 ,X 2 ,···X i ,X j ,···X n , Where n rows represent the number of data points collected in one door opening and closing process, which is 3 in this embodiment; column p represents that the extracted features are 5 in this embodiment.
[0049] Describe the mean value of data extraction, the expression is as follows:
[0050]
[0051] Calculate the maximum and minimum values ​​of the door-opening and door-closing processes of speed-up, speed-up, speed-down, slow-moving section, and post-positioning section: reflect the range of data changes.
[0052] X max =max|x i |
[0053] X min =min|x i |
[0054] Calculate the variances of the door-opening and door-closing processes of increasing speed, high speed, slowing speed, slow-moving section, and post-positioning section: Describe the degree of deviation of the data from the mean value.
[0055]
[0056] Calculate the skewness of the door-opening and door-closing processes of speed-up, high-speed, slow-down, slow-moving and post-position: a measure reflecting the skewness of data distribution, Skewness> 0 is called right skewness. At this time, the distribution trend of the data is that most of the data are on the right side of the mean; Skewness <0 is called left skewness, and the skewness of the data distribution is the opposite; when Skewness is close to 0, the data distribution can be considered symmetrical.
[0057]
[0058] Calculate the kurtosis of the door-opening and door-closing processes of the ascending, high-speed, decelerating, slow-moving section and the post-in-place section: reflect the peak level of the probability density distribution of the data within the average range. Normally distributed data exhibits kurtosis of 3. If the skewness of the sample data is much greater than 3, it means that the peak of the sample data is relatively steep, indicating that there are more data in the sample data far away from the mean, so generally kurtosis can be used to measure the sample data The degree of deviation from the normal distribution.
[0059]
[0060] X in the above formula i Is the value of the system state variable.
[0061] In frequency domain feature extraction, the system feature set is decomposed into multiple independent frequency domain subspaces, and the wavelet decomposition method is used to extract the energy frequency domain energy features of each frequency band.
[0062] Wavelet transform is a local transformation method of time and frequency. It is improved on the basis of the short-time local transformation of Fourier transform. It can change the size of the window according to the change of frequency. It has been successfully applied in many fields, such as signal processing. , Image processing and pattern recognition, etc. An important feature of wavelet change is that it has good localization characteristics in the frequency domain. In wavelet transform, two functions must be defined first.
[0063]
[0064] among them, Is the scale function, and ψ(t) is the wavelet function.
[0065] Signal x(t)∈L 2 (R) The decomposition signal in one of the wavelet sub-vector spaces is
[0066]
[0067] Where Is the node, the wavelet packet coefficient corresponding to (j,n).
[0068] Then the wavelet decomposition of signal x(t) can be written as follows:
[0069]
[0070] Where ω n,j,k (t) is the orthogonal wavelet basis.
[0071] Sub-band signal The energy of is calculated by the following formula:
[0072]
[0073] When analyzing the data of the rail vehicle door system in the present invention, the high-frequency components of the data are no longer decomposed, but the low-frequency components of the data are continued to be decomposed. Choose a three-layer wavelet decomposition structure here. The position, speed and current signals of door opening and closing collected by the motor of the rail vehicle door system are respectively subjected to wavelet decomposition, and the energies E1, E2, E3 and E4 of 4 sub-bands are obtained as the sub-healthy frequency domain energy characteristics.
[0074] (C) Multi-door system health modeling, including the following steps,
[0075] (C1) A "decision diagram" through the relative distance and local density of multiple door features, such as figure 2 As shown, the density peak is manually selected, that is, the cluster center. The ideal cluster center (density peaks) has two basic characteristics: the local density ρ of sample i i Greater than the local density of its neighbors and the distance δ between the centers j of different clusters i Relatively far away. Its definition is as follows,
[0076]
[0077]
[0078] Where d ij Is the data point X after feature extraction i ,X j Euclidean distance between d c Is the cutoff distance (set manually). For local density ρ i Largest sample X i , Where δ i =max j d ij.
[0079] (C2) Assign the remaining data points to obtain the clustering result. For the remaining system state variable value X j , Put it into the density ratio X j Large and distance X j The most recent system state variable value belongs to the cluster, and the allocation of the remaining system state variable value j is completed in one step.
[0080] Step (D): Perform periodic horizontal comparison based on Euclidean distance on the obtained vehicle multi-door model, including: After finding the cluster center, calculate the Euclidean distance from each door to the density center, such as image 3 Shown, and then calculate the mean, such as Figure 4 , Figure 5 Shown. The door with the farthest distance from the center of density every day is defined as an "abnormal door" and marked. In specific embodiments, normal doors can be marked at the same time to identify the health status of the door and complete the door health status based on the normal operation data Automatic identification and assignment process of health labels.
[0081] After this embodiment, the statistical assignment results are sorted and compared and analyzed with historical fault information, which verifies the effectiveness of the algorithm. The invention adopts a data-based mathematical modeling method, uses a density peak clustering algorithm to perform mathematical modeling, and establishes a health model of the multi-vehicle door system; the established model is used to periodically perform periodicity on the acquired multi-vehicle door model It completes the abnormal detection of the vehicle multi-door system and reduces the repetitive experimental design and data collection work. It is the first application of the density peak clustering algorithm in the rail vehicle door fault detection technology, regardless of the door type or locking The influence of device type has good universality.
[0082] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principles of the present invention, several improvements and modifications can be made. These improvements and modifications It should also be regarded as the protection scope of the present invention.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Similar technology patents

Method allowing licensed pharmacist to use mobile phone or tablet computer to remotely check prescription

InactiveCN106682378AReduce duplicationCause blockageComputer-assisted medicine prescription/deliverySpecial data processing applicationsTablet computerDrug product
Owner:SHANGHAI HYDEE SOFTWARE CORP

Classification and recommendation of technical efficacy words

  • Reduce duplication
  • Universal

Voice identification method and device of mobile terminal

ActiveCN103280217AAdapt to the characteristics of weak processing capacityReduce duplicationDevices with voice recognitionSubstation equipmentSpeech identificationIdentification device
Owner:BEIJING BYTEDANCE NETWORK TECH CO LTD

Single-linear-array laser radar equipment centering method

ActiveCN104483664AInhibiting requirements for high-precision data acquisition and validationUniversalWave based measurement systemsLinear arrayEnvironmental geology
Owner:ACAD OF OPTO ELECTRONICS CHINESE ACAD OF SCI
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products