CN112816052B - Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method - Google Patents

Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method Download PDF

Info

Publication number
CN112816052B
CN112816052B CN202110003658.8A CN202110003658A CN112816052B CN 112816052 B CN112816052 B CN 112816052B CN 202110003658 A CN202110003658 A CN 202110003658A CN 112816052 B CN112816052 B CN 112816052B
Authority
CN
China
Prior art keywords
vibration
measuring point
train
signal
total measuring
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110003658.8A
Other languages
Chinese (zh)
Other versions
CN112816052A (en
Inventor
刘辉
于程名
李燕飞
秦进
张雷
尹诗
段铸
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Central South University
Original Assignee
Central South University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Central South University filed Critical Central South University
Priority to CN202110003658.8A priority Critical patent/CN112816052B/en
Publication of CN112816052A publication Critical patent/CN112816052A/en
Application granted granted Critical
Publication of CN112816052B publication Critical patent/CN112816052B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G01H1/04—Measuring characteristics of vibrations in solids by using direct conduction to the detector of vibrations which are transverse to direction of propagation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G01H1/04—Measuring characteristics of vibrations in solids by using direct conduction to the detector of vibrations which are transverse to direction of propagation
    • G01H1/08—Amplitude
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G01H1/12—Measuring characteristics of vibrations in solids by using direct conduction to the detector of longitudinal or not specified vibrations
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
    • G01H1/12—Measuring characteristics of vibrations in solids by using direct conduction to the detector of longitudinal or not specified vibrations
    • G01H1/16—Amplitude
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M17/00—Testing of vehicles
    • G01M17/08—Railway vehicles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Automation & Control Theory (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

The invention discloses a train compartment vibration monitoring method, a vibration signal characteristic library establishment and application method, wherein a non-invasive train compartment vibration monitoring method is adopted, only vibration signals of a total measuring point in a train compartment need to be acquired during actual monitoring, sensor redundancy is avoided, cost is saved, the train is beneficial to light weight, the relation between the position of a vibration source and phase deviation and amplitude attenuation is accurately and truly reflected, the vibration condition of an unknown vibration source can be monitored, abnormal vibration of the vibration source can be identified, and thus a basis is provided for abnormal vibration processing; the reconstruction of the vibration source signal can be realized on the basis of only measuring the vibration signal of the total measuring point, so that the real-time update of the multivariate vibration signal feature library is realized, and the multivariate vibration signal feature library can provide a data basis for the optimized decomposition of the vibration signal, the image identification of the vibration signal, the identification of the abnormal vibration signal and the like.

Description

Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method
Technical Field
The invention belongs to the technical field of train room vibration identification, and particularly relates to a train room vibration monitoring method, and a vibration signal characteristic library establishment and application method.
Background
In recent years, high-speed trains in China are continuously developed, in the high-speed running process of the trains, the interior of a train room can be influenced by vibration from the outside to generate vibration, and the vibration of the train room can reflect the running condition of train parts in contact with the train room on one hand and influence the comfort and riding experience of passengers on the other hand.
The existing method for monitoring the vibration of the train compartment in real time has the following defects:
firstly, the sensor is only mounted at the key position, and the monitoring of the unknown vibration source cannot be realized. For example, patent publication No. CN111044303A proposes a method for diagnosing abnormal vibration of passenger compartment of magnetic levitation train, which installs a large number of sensors in the passenger compartment of the magnetic levitation train to monitor the abnormal vibration of the passenger compartment of the magnetic levitation train, but the patent does not relate to monitoring unknown vibration source.
Secondly, the current monitoring mode is mainly to monitor a large number of sensors installed in a intrusive mode on the train directly, and although the accuracy is high, the situation of sensor redundancy and waste is inevitably generated. For example, patent publication No. CN110879102A proposes a vibration monitoring system for rail trains, which is configured to set multiple vibration monitoring terminals in each car and transmit data to a master control center in real time, so as to determine whether the car vibration is abnormal, but this method does not involve monitoring the train vibration source, and requires more monitoring devices such as sensors, and thus has higher cost.
Based on the reason, the non-invasive vibration monitoring is carried out on the train room, and the source of the abnormal vibration is determined, so that the method has important significance for reducing the use and maintenance cost of the train, ensuring the operation safety of the train and improving the comfort of passengers.
Disclosure of Invention
The invention aims to provide a train compartment non-invasive vibration monitoring method, a vibration signal characteristic library establishment method and an application method aiming at the defects of the prior art, a large number of sensors are not needed, and the abnormal vibration monitoring of the known vibration source and the unknown vibration source of a train can be realized.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a train compartment non-invasive vibration monitoring method is characterized by comprising the following steps:
step 1, respectively pre-collecting vibration data M of a plurality of branch measuring points in the train room in the transverse direction, the longitudinal direction and the vertical direction under the condition of normal vibration and abnormal vibration of a known vibration source outside the train room, pre-collecting vibration data C of a total measuring point in the train room in the transverse direction, the longitudinal direction and the vertical direction, and pre-collecting vibration data S of each known vibration source outside the train room in the transverse direction, the longitudinal direction and the vertical direction;
step 2, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data M of each branch measuring point, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data C of the total measuring point, and extracting phase information and amplitude information of first 1-J harmonic signals of each known vibration source;
step 3, taking the first 1-J harmonic phase difference and the harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the position information of each non-total measuring point relative to the total measuring point in the transverse direction, the longitudinal direction and the vertical direction as output, applying deep echo state network training, adopting the deep echo state network to construct a model for the harmonic phase difference and the harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point and the position information of each non-total measuring point relative to the total measuring point, wherein the model is a model I,
the parameters of the deep echo state network are set as follows, the number of the nodes of the storage pool is set to be 10, 10-fold cross validation is adopted to perform self-adaptive determination on the number of layers of the storage pool and the radius of a matrix spectrum, and the selection range is [1,2, 3., 10]And [0.1,0.2,..., 0.9 ]]Selecting parameters capable of accurately reflecting the relationship, and constructing a nonlinear relationship model f of harmonic phase difference and harmonic amplitude ratio of each non-total measuring point and position information of each non-total measuring point relative to the total measuring point1z(I)，f1h(I)，f1c(I) Respectively denote longitudinal, transverse and verticalA vibration signal relation model;
taking the first 1-J harmonic phase difference between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the first 1-J harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as output, training by using a deep echo state network, constructing a model by using the deep echo state network on the relation between the harmonic phase difference and the harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point signal, wherein the model is a model II,
the parameters of the deep echo state network are set as follows, the number of the nodes of the storage pool is set to be 10, 10-fold cross validation is adopted to carry out self-adaptive determination on the number of layers of the storage pool and the radius of a matrix spectrum, and the selection range is [1,2, 3.., 10., [1 ]]And [0.1,0.2,..., 0.9 ]]Selecting parameters capable of accurately reflecting the relation, and constructing a relation model f of the harmonic phase shift and the harmonic amplitude ratio of each vibration signal of the non-total measuring points relative to the vibration signal of the total measuring points in the longitudinal direction, the transverse direction and the vertical direction2z(I)，f2h(I)，f2c(I) Respectively representing a longitudinal vibration signal relation model, a transverse vibration signal relation model and a vertical vibration signal relation model,
the non-total measuring point consists of a measuring point and a known vibration source;
step 4, collecting real-time vibration data CR of the total measuring point in the transverse direction, the longitudinal direction and the vertical direction;
and 5, setting an optimization target by taking the phase offset of the vibration source relative to the total measuring point, the combined working state of each known vibration source and the abnormal vibration type as independent variables, executing a multi-objective optimization algorithm based on the first model and the second model, and outputting vibration monitoring result information of the train compartment, wherein the vibration monitoring result information comprises the position of the vibration source generating vibration and/or the abnormal vibration type of the vibration source generating vibration and/or the first 1-J harmonic phase offset of the vibration source generating vibration relative to the total measuring point and/or the first 1-J harmonic amplitude ratio of the vibration source generating vibration relative to the total measuring point.
In a preferable mode, in the step 1, the normal vibration of the known vibration source outside the train room includes two situations: the first type is that each known vibration source outside the train room normally vibrates, and the second type is that any known vibration source outside the train room independently normally vibrates and the other known vibration sources do not vibrate.
In a preferable mode, in the step 1, the abnormal vibration of the known vibration source outside the train room includes two situations: the first type is that any known vibration source outside the train room vibrates independently and the other known vibration sources do not vibrate, and the second type is that any known vibration source outside the train room vibrates independently and the other known vibration sources vibrate normally.
In a preferred embodiment, in step 1, M ═ L is providedM,m(t)]，C＝[LC,c(t)]， S＝[LS,s(t)](ii) a m (t) is a vibration signal value corresponding to the branch measuring point, c (t) is a vibration signal value corresponding to the total measuring point, and s (t) is a vibration signal value corresponding to the known vibration source; l isMIs a label value L of the vibration data corresponding to the measured pointM＝[lM,lt,lg,ld,lo1]，LCA label value L of the vibration data corresponding to the total measuring pointC＝[ls,lt,lg,ld]，LSTag value L of vibration data corresponding to known vibration sourceS＝[lS,lt,lg,ld,lo2]，lMNumber labels for the distribution points, /)tIs a time stamp tag,/gFor abnormal vibration type label,/o1Is a position information label of the measuring point relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction,/sFor known vibration source tags which vibrate simultaneously,/dFor vibrating direction labels, /)SNumber label for known vibration source,/o2The position information labels of the known vibration source relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction are obtained.
Preferably, in the step 5, minimizing a weighted difference between the sum of the frequency domain characteristics of the vibration signals of the known vibration sources and the frequency domain characteristics of the harmonic orders of the vibration signals of the total measuring point as one of the optimization objectives; so as to minimize the weighted variance of the harmonic order differences of the vibration signals of the known vibration sources to be the second optimization objective.
Further, the method also comprises a step 6 of converting and restoring the vibration monitoring result information obtained in the step 5 into a real-time vibration source vibration signal in a time domain.
Based on the same inventive concept, the invention also provides a method for establishing a train compartment vibration signal characteristic library, which is characterized by comprising the following steps:
step 7, extracting frequency domain characteristics, time domain characteristics and image characteristics of the real-time vibration source vibration signal, and performing characteristic fusion or characteristic reconstruction on the frequency domain characteristics and the time domain characteristics of the real-time vibration source vibration signal;
and 8, establishing a train compartment vibration signal feature library based on the frequency domain feature, the time domain feature, the image feature, the feature fusion or the feature reconstruction result obtained in the step 7.
And further, performing online training on the model I and/or the model II by using information in a train compartment vibration signal feature library.
Based on the same inventive concept, the invention also provides an application method of the train compartment vibration signal characteristic library established by the method for establishing the train compartment vibration signal characteristic library, which is characterized by comprising the following steps of:
training to obtain a relation model of vibration signal characteristics in a train compartment vibration signal characteristic library and train service performance characterization parameters, and outputting to obtain real-time train service performance characterization parameters by taking the vibration signal characteristics in the train compartment vibration signal characteristic library obtained in real time as the input of the relation model in the running process of a train.
Based on the same inventive concept, the invention also provides an application method of the train compartment vibration signal characteristic library established by the method for establishing the train compartment vibration signal characteristic library, which is characterized by comprising the following steps of:
and predicting the abnormal vibration condition of the vibration source at a certain moment in the future by using the time series information of the vibration signal characteristics in the train compartment vibration signal characteristic library and a time series prediction method.
Compared with the prior art, the invention has the following beneficial effects:
1) by adopting a non-invasive train compartment vibration monitoring method, only vibration signals of a total measuring point in a train compartment need to be acquired during actual monitoring, so that sensor redundancy is avoided, cost is saved, and light weight of a train is facilitated.
2) The modeling is carried out on the relation between the harmonic phase offset and the harmonic amplitude attenuation of the vibration signal of each branch measuring point relative to the vibration signal of the total measuring point and the relative position information, the nonlinear modeling of the relation can be realized, the relation between the vibration source position and the phase offset and the amplitude attenuation can be accurately and really reflected, the foundation is laid for the non-invasive vibration monitoring, and the vibration condition of an unknown vibration source can be monitored.
3) A multi-objective optimization method is adopted to determine the optimal phase shift and the optimal vibration signal combination, and further the vibration source position calculation is realized through the relation between the phase shift and the position information. The method only needs to measure the vibration signal of the total measuring point during actual monitoring, and can identify the abnormal vibration of the vibration source, thereby providing a basis for processing the abnormal vibration.
4) The method can realize the reconstruction of the vibration source signal on the basis of only measuring the vibration signal of a total measuring point, thereby realizing the real-time update of the multivariate vibration signal feature library, and the multivariate vibration signal feature library can provide a data basis for the optimized decomposition of the vibration signal, the image identification of the vibration signal, the identification of the abnormal vibration signal and the like.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
Detailed Description
In the high-speed running process of a train, the interior of a train cabin can be influenced by an external vibration source to generate vibration, the vibration of the train cabin can reflect the running condition of train parts in contact with the train cabin on one hand, the comfort of passengers and riding experience can be influenced on the other hand, non-invasive monitoring is carried out on the vibration of the train cabin, important support can be provided for determining the service performance and the abnormal vibration source of the train, and the method has important significance for guaranteeing the running safety of the train and improving the comfort of the passengers.
As shown in fig. 1, the present invention comprises the steps of:
step 1, invasive pre-acquisition of vibration data of train compartments at multiple acquisition points
Before non-invasive vibration monitoring of a train compartment, part of initial training data needs to be collected in advance. The initial training data is acquired by adopting multi-measuring point invasive type:
in step 101, a plurality of branch measuring points and a total measuring point are arranged in a train compartment. Wherein, the arrangement rule of the distributed points is as follows: aiming at a plurality of transverse sections of a train, arranging at least two vibration sensors for measuring transverse vibration signals of a train compartment on each transverse section; aiming at a plurality of longitudinal sections of a train, arranging at least two vibration sensors for measuring longitudinal vibration signals of a train compartment on each longitudinal section; aiming at a plurality of vertical sections of the train, at least two vibration sensors for measuring vertical vibration signals of a train compartment are arranged on each vertical section. And 3 vibration sensors are arranged at the total measuring point and are respectively used for measuring transverse, longitudinal and vertical vibration signals at the total measuring point. Meanwhile, aiming at each known vibration source outside the train compartment, 3 vibration sensors are arranged at each known vibration source and are respectively used for measuring transverse, longitudinal and vertical vibration signals at the corresponding known vibration source.
102, acquiring vibration signals of all branch measuring points, vibration signals of a total measuring point and vibration signals of an external known vibration source of a train room in the normal running process of the experimental train (namely all the known vibration sources outside the train room vibrate normally).
103, designing an independent vibration experiment of each vibration source (namely, any known vibration source outside the train room vibrates independently and normally, and the other known vibration sources do not vibrate) and an abnormal vibration experiment of each vibration source, wherein the abnormal vibration experiment of each vibration source is divided into an independent vibration experiment of each abnormal vibration of each vibration source (namely, any known vibration source outside the train room vibrates independently and the other known vibration sources do not vibrate) and a mixed vibration experiment of single abnormal vibration source and multiple vibration sources (namely, any known vibration source outside the train room vibrates independently and normally, and the vibration signals of each measuring point, the vibration signals of the total measuring points and the vibration signals of the external known vibration sources of the train room are collected, and the same experiment is repeated for five times.
Step 104, storing the vibration signals obtained in step 102 and step 103 in a data storage module in real time, wherein the following data storage format is mainly adopted: vibration data M ═ L of a plurality of branch measuring points in the transverse direction, the longitudinal direction and the vertical direction in the train roomM,m(t)]Vibration data C ═ L of train indoor total measuring point in transverse direction, longitudinal direction and vertical directionC,c(t)]Vibration data S ═ L of all known vibration sources outside the train room in the transverse direction, the longitudinal direction and the vertical directionS,s(t)](ii) a m (t) is a vibration signal value corresponding to the branch measuring point, c (t) is a vibration signal value corresponding to the total measuring point, and s (t) is a vibration signal value corresponding to the known vibration source; l isMIs a label value L of the vibration data corresponding to the measured pointM＝[lM,lt,lg,ld,lo1]，LCA label value L of the vibration data corresponding to the total measuring pointC＝[ls,lt,lg,ld]，LSTag value L of vibration data corresponding to known vibration sourceS＝[lS,lt,lg,ld,lo2]，lMNumber labels for the points of distribution (for distinguishing signals from different stations in the vehicle), ltIs a time stamp tag,/gFor abnormal vibration type label (abnormal vibration type is determined by the prior art), lo1Is a position information label of the measuring point relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction,/sFor known vibration source tags that vibrate simultaneously (for marking the vibration source that is currently vibrating), ldFor vibration orientation tags (for distinguishing transverse, longitudinal or vertical vibration signals), lSNumber label for known vibration source (for distinguishing signals from different vibration source measuring points), lo2The position information labels of the known vibration source relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction are obtained.
The tag value is automatically given according to the ID of the vibration sensor which transmits the signal back, and the abnormal type tag needs to be manually marked, but can be automatically given after non-invasive characteristic collection is realized.
And 2, extracting frequency domain characteristics of the vibration signals of the vehicle room, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data M of each branch measuring point, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data C of the total measuring point, and extracting phase information and amplitude information of first 1-J harmonic signals of each known vibration source. The method comprises the following specific steps:
in order to further analyze the signals and realize non-invasive monitoring of the signals, the frequency domain characteristics of the signals of the initial training set need to be analyzed and extracted. The time-frequency domain characteristics of the vibration signals generated by different vibration sources and the characteristics of harmonic phase shift, signal time shift, amplitude attenuation and the like of the vibration signals and a total vibration signal measuring point are distinguishable; the signal characteristics corresponding to different abnormal vibration conditions of the same vibration source are unique, and the accuracy of non-invasive monitoring and characteristic acquisition can be improved by analyzing the time domain and the frequency domain of the signal.
FFT (fast Fourier transform) is carried out on the signal values m (t), c (t) and s (t) of the vibration signals of the general measuring point, the measuring points and the vibration source of the train room measured in the step 1, the vibration time sequence signal for frequency domain analysis is 5s, namely, the frequency domain analysis is carried out on each sample signal at the interval of 5s, and 1-J order harmonic signals { m & lt m & gt of each vibration signal can be obtained1(t),m2(t),…,mJ(t)}、{c1(t),c2(t),…,cJ(t)}、{s1(t),s2(t),…,sJ(t) and obtaining amplitude information and phase information of each harmonic order of the vibration signal, namely mA ═ mA }1(t),ma2(t),…, maJ(t)}、cA＝{ca1(t),…,ca2(t),caJ(t)}、sA＝{sa1(t),sa2(t),…,saJ(t) } and
Figure GDA0003362939930000071
Figure GDA0003362939930000072
wherein m isJThe J-th harmonic signal, s, representing the signal at the point of measurement of the vehicle cabinJThe J-th harmonic signal, c, representing the vibration source signalJA J-th harmonic signal representing a source signal, andaffixes A and B represent the magnitude and phase sets, respectively, suffixes a and
Figure GDA0003362939930000073
representing amplitude and phase values, respectively.
And 3, modeling the relation data of harmonic phase shift, amplitude attenuation and measuring point position of the vibration signal of the vehicle room.
In the step 3, taking the first 1-J harmonic phase difference and the harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the position information of each non-total measuring point relative to the total measuring point in the transverse direction, the longitudinal direction and the vertical direction as output, and training by using a machine learning algorithm to obtain a model I; taking the first 1-J harmonic phase difference between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the first 1-J harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as output, and training by using a machine learning algorithm to obtain a second model; wherein, the non-total measuring point is composed of a measuring point and a known vibration source.
The method comprises the following specific steps:
the invention adopts a data modeling mode to analyze the relationship between the phase shift and the amplitude attenuation of vibration signals of different measuring points and the positions of the measuring points. In an ideal state, the phase offset and amplitude attenuation of the vibration signal have a definite relation with the signal propagation distance, but due to the complex structure of train compartment materials and the like, the traditional method can not reflect the real relation between the signal offset and amplitude attenuation and the position of a measuring point, so the method adopts a data modeling mode to analyze the vibration signal.
Step 301, calculating the harmonic phase difference between the signal of the non-total measuring point (including the vibration signal of the vehicle compartment measuring point and the known vibration source signal) and the vibration signal of the total measuring point
Figure GDA0003362939930000081
Harmonic amplitude ratio ukiAnd the position relation l of the non-measuring point signal and the total measuring point signaloWherein k represents a signal k, i represents a harmonic order, and the harmonic phase differences of the measuring points measured for multiple times are averaged to obtain the measuring points or the known vibration source relative to the total measuring pointHarmonic phase difference and amplitude ratio of harmonic orders of vibration signal
Figure GDA0003362939930000082
Step 302, modeling harmonic phase difference, harmonic amplitude ratio-total measuring point relative position relation data
(1) And reading harmonic phase difference and harmonic amplitude ratio data of the transverse, longitudinal and vertical branch measuring points and the vibration source measuring points relative to the total measuring point and relative total measuring point position relation data, wherein 80% of a data set is used as a training set, and 20% of the data set is used as a test set.
(2) Taking the harmonic phase difference and the harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point signal as input,
Figure GDA0003362939930000083
I＝[I1,I2,…Ik]taking the position of each non-total measuring point relative to the total measuring point as output O ═ lo1,lo2,…lok]。
(3) Modeling and describing harmonic phase difference and harmonic amplitude ratio of each non-total measuring point signal relative to a total measuring point signal and position information of each non-total measuring point relative to a total measuring point by adopting a deep echo state network, setting parameters of the deep echo state network as follows, setting the number of nodes of a storage pool as 10, and carrying out self-adaptive determination on the number of layers of the storage pool and the matrix spectrum radius by adopting 10-fold cross validation, wherein the selection range is [1,2,3, …,10 ]]And [0.1,0.2, …,0.9 ]]Selecting parameters capable of accurately reflecting the relationship, and finally constructing a nonlinear relationship model f of harmonic phase difference and harmonic amplitude ratio of each non-total measuring point and position information of each non-total measuring point relative to the total measuring point1z(I),f1h(I),f1c(I) And respectively represent a longitudinal vibration signal relation model, a transverse vibration signal relation model and a vertical vibration signal relation model.
Step 303, modeling the relation data of harmonic phase difference and harmonic amplitude ratio of signals of each non-total measuring point
(1) And reading harmonic phase difference and harmonic amplitude ratio data of the transverse, longitudinal and vertical branch measuring points and the vibration source measuring points relative to the total measuring point, wherein 80% of a data set is used as a training set, and 20% of the data set is used as a test set.
(2) Taking the harmonic phase difference of each non-total measuring point signal relative to the total measuring point signal as input,
Figure GDA0003362939930000084
I＝[I1,I2,…Ik]and taking the harmonic amplitude ratio of each non-total measuring point to the total measuring point as output O ═ u ═1,u2,…uk]。
(3) Modeling and describing the relation between harmonic phase difference and harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point signal by adopting a deep echo state network, setting the parameters of the deep echo state network as follows, setting the number of nodes of a storage pool to be 10, and carrying out self-adaptive determination on the number of layers of the storage pool and the matrix spectrum radius by adopting 10-fold cross validation, wherein the selection range is [1,2,3, …,10 ]]And [0.1,0.2, …,0.9 ]]Selecting parameters capable of accurately reflecting the relation, and finally constructing a relation model f of the harmonic phase shift and the harmonic amplitude ratio of each vibration signal of the non-total measuring points relative to the vibration signal of the total measuring points in the longitudinal direction, the transverse direction and the vertical direction2z(I),f2h(I),f2c(I) And respectively represent a longitudinal vibration signal relation model, a transverse vibration signal relation model and a vertical vibration signal relation model.
And 304, repeating the steps 302 to 303 for J times according to the harmonic order to obtain a corresponding relation model of each harmonic order of the vibration signal.
Step 4, non-invasive collection of transverse, longitudinal and vertical real-time vibration data CR of train room total measuring point
In the high-speed running process of a train, a total measuring point vibration sensor in the transverse direction, the longitudinal direction and the vertical direction is only arranged at a total measuring point in a train room, vibration signals of the total measuring point in the three directions are collected in real time, a timestamp is recorded, the real-time collected data are sent to a data storage module in a 4G mode, and the data transmission interval is 1 minute.
Step 5, the non-invasive vibration source signal monitoring and phase shift determination step 5 based on multi-objective optimized phasor decomposition specifically comprises the following steps:
step 501, selecting a multi-objective gray wolf optimization method to construct a vibration source signal and a phase shift determination model thereof, selecting a multi-objective optimization method and reasonably selecting corresponding hyper-parameters: and (3) determining corresponding parameters such as the maximum iteration number of 50, the grey wolf number of 100 and the stock number of 50 by adopting a multi-objective particle swarm optimization model. The model feature library only adopts a feature library formed by frequency domain features of vibration signals of the vibration source.
In step 502, the optimization variables are different vibration sources and abnormal vibration types.
(1) The independent variables mainly include: phase deviation of the vibration source relative to a total measuring point, combined working state of each known vibration source and abnormal vibration type, wherein the phase deviation of the vibration source relative to the total measuring point is a continuous independent variable and is used for optimizing proper phase, the combined working state of each known vibration source and the abnormal vibration type are discrete independent variables and are used for optimizing proper vibration source signal source and abnormal vibration type, and the independent variables can be expressed as signal sources of the vibration source and the abnormal vibration type
Figure GDA0003362939930000091
Wherein D is used for limiting the number of vibration sources,
Figure GDA0003362939930000092
representing a phase shift;
(2) corresponding continuous independent variable is
Figure GDA0003362939930000093
Where d is [0.5, z +0.5) ], Lci∈(0,1),Lgi∈(0,1)，
And D ═ D +1/2]，lci＝[Lci]，lgi＝[Lgi]I.e. D is rounded up by D, lci、lgiAre respectively Lci、LgiRounding off of (d), lci、lgiThere is no other special meaning.
Step 503, according to the phase shift in the independent variable, using the relation model of the harmonic phase shift and the harmonic amplitude ratio in the longitudinal direction, the transverse direction and the vertical direction of the vibration signal of each non-total measuring point relative to the vibration signal of the total measuring point obtained in step 303Type f2z(I),f2h(I),f2c(I) Calculating the harmonic amplitude ratio u of the total measuring point signal part corresponding to each vibration source signaliFurther obtain the harmonic phasor corresponding to each vibration source signal
Figure GDA0003362939930000101
Figure GDA0003362939930000102
Wherein
Figure GDA0003362939930000103
And j harmonic phasor of a vibration signal generated by the ith vibration source signal at the total measuring point in the pre-acquired data characteristic library is represented, and the phasor is a vector consisting of the transformation ratio amplitude, the phase and the phase offset of the vibration source signal.
Step 504, setting an optimization target to minimize a weighted difference between the sum of frequency domain characteristics of the vibration signals of the known vibration sources and the frequency domain characteristics of harmonic orders of the vibration signals of the total measuring point as one of the optimization targets; so as to minimize the weighted variance of the harmonic order differences of the vibration signals of the known vibration sources to be the second optimization objective. The optimization objective function is as follows:
Figure GDA0003362939930000104
Figure GDA0003362939930000105
wherein, F1And F2Representing two objective functions, wjIs the weight of each order of the harmonics,
Figure GDA0003362939930000106
representing the j harmonic phasor value of the real-time total measurement point signal, wherein the phasor consists of the corresponding amplitude and phase; a is the total number of known vibration sources.
And 505, performing multi-objective grey wolf optimization on the independent variables, calculating the optimization function values of all search results, and selecting a non-dominant solution to store in the file.
Step 506, updating the search path to generate a new argument scheme.
Step 507, if It is less than the maximum iteration number, returning to step 504; otherwise, the multi-objective optimization algorithm is ended, and the non-dominated solution set NS in the final archive is output.
Step 508, the argument in the non-dominated solution set that minimizes the objective function is taken as the final solution.
Step 509, according to the selected optimal vibration source combination and the optimal phase shift, executing step 503 to obtain a corresponding harmonic amplitude ratio; with the optimal phase shift and the corresponding harmonic amplitude ratio as input, applying the harmonic phase difference, the harmonic amplitude ratio and the relationship model f between the signals of the non-total measuring points and the total measuring points trained in step 4021z(I),f1h(I),f1c(I) The location of the vibration source where vibration is currently occurring is obtained.
And step 5010, outputting vibration monitoring result information of the train compartment, wherein the vibration monitoring result information comprises the position of a vibration source generating vibration and/or the abnormal vibration type of the vibration source generating vibration and/or the first 1-J harmonic phase deviation of the vibration source generating vibration relative to the total measuring point and/or the first 1-J harmonic amplitude ratio of the vibration source generating vibration relative to the total measuring point.
Step 5011, taking corresponding vibration reduction measures according to the obtained abnormal vibration position to meet the comfort requirement of passengers; and corresponding fault early warning measures are taken to avoid faults.
And 6, restoring the real-time non-invasive vibration source signal time domain, namely converting the vibration monitoring result information obtained in the step 5 into a real-time vibration source vibration signal in the time domain.
And 601, based on the optimization result of the step 5, reducing each vibration source signal by utilizing harmonic phase shift and harmonic amplitude comparison. First according to the optimum result obtained in step 5c1,lc2… …), extracting a frequency domain signal f of a vibration source corresponding to the pre-acquisition data feature library, and correcting the pre-acquisition vibration source signal according to the following formula to obtain a real-time vibration source frequency domain signal:
Figure GDA0003362939930000111
wherein A issrAnd
Figure GDA0003362939930000112
representing the frequency domain amplitude and harmonic phase, A, of a real-time signalsoAnd
Figure GDA0003362939930000113
representing the frequency domain amplitude and harmonic phase u of the pre-acquired vibration source signalsoAnd
Figure GDA0003362939930000114
representing the harmonic amplitude ratio and the harmonic phase shift u of the pre-acquired vibration source signal relative to the total measuring pointsrAnd
Figure GDA0003362939930000115
and expressing the optimal harmonic phase shift and harmonic amplitude ratio obtained by optimization. In the process of frequency domain signal correction, the harmonic amplitude ratio corresponding to each harmonic is adopted for amplitude correction near each harmonic. Finally, a real-time vibration source signal s (f) can be obtained.
Step 602, inverse fourier transform is performed on the frequency domain value of the real-time vibration signal by using an IFFT technique to obtain a time domain real-time vibration source signal, where s (t) is IFFT (s (f)).
Step 7, extracting the real-time non-invasive vibration source signal characteristics and fusing or reconstructing the characteristics
Step 701: real-time vibration signal feature extraction
(1) Extracting signal frequency domain features: and (4) repeating the frequency domain feature extraction method in the step (2) aiming at the real-time vibration source signal obtained in the step (6) to extract the frequency domain feature of the real-time vibration source signal.
(2) Extracting signal time domain features: for the real-time vibration source signal, time domain features of the real-time vibration source signal are extracted, such as maximum amplitude, time offset, 5s mean value and the like.
(3) Extracting the characteristics of the signal image: and performing feature extraction on the time domain and frequency domain images of the real-time vibration source signal to obtain image features.
Step 702: real-time vibration signal feature fusion
And performing feature fusion or feature reconstruction on the time domain and frequency domain features by adopting a KPCA (kernel principal component analysis) method, wherein a kernel function adopts a Gaussian kernel, the accumulated contribution rate of the features is calculated in the feature fusion process, and the features with the accumulated contribution rate of more than 95 percent are taken as a feature fusion result. All feature data were cross-validated by 10 folds to determine the optimal feature count.
Step 8, establishing a non-invasive real-time updated train room vibration signal multivariate feature library
Establishing a multivariate feature library based on the signal features extracted in the step 7 according to the feature application method and the feature characteristics:
(1) establishing a non-invasive vibration signal optimization decomposition characteristic library
In the real-time vibration source signal characteristics obtained in the above steps, the frequency domain characteristics include characteristics such as harmonic phasor, harmonic amplitude, harmonic phase shift, harmonic amplitude ratio and the like, so that the vibration source signal frequency domain characteristic library can be further enriched, and the frequency domain characteristics are stored in the data storage module according to the data format of the step 2, so that the real-time updating and supplementing of the non-invasive vibration signal multi-objective optimization decomposition characteristic library are completed. The library of features can be re-applied to non-invasive vibration monitoring.
(2) Establishing a non-invasive vibration signal identification feature library
The non-invasive vibration signal identification feature library updated in real time can be established according to the real-time vibration source signal features of different positions obtained in the steps, including frequency domain features, time domain features and fusion features of the real-time vibration source signal features, and can be applied to identification of vibration signals of different sources by an artificial intelligence algorithm.
(3) Establishing a non-invasive abnormal vibration signal identification feature library
According to the different real-time vibration source signal characteristics obtained in the steps, the abnormal vibration signal characteristics are appointed and collected, and a non-invasive abnormal vibration signal identification characteristic library updated in real time can be established.
(4) Establishing a non-invasive real-time updated train compartment vibration image recognition feature library
The deep learning algorithm based on image learning can more accurately identify vibration characteristics, and a huge train room vibration source image characteristic library can be constructed according to vibration source signals collected in a non-invasive manner in real time, so that a foundation is laid for the application of an artificial intelligent image identification method.
(5) Establishing a non-invasive real-time updated train compartment vibration fusion characteristic and reconstruction characteristic database
The fusion characteristics of each vibration source signal can be obtained according to the characteristic fusion and characteristic reconstruction method in the step 7, the fusion characteristics have more obvious effect in the vibration signal identification process, and the establishment of the non-invasive train compartment vibration signal fusion characteristic library is helpful for promoting the application of an artificial intelligence algorithm in the field.
Step 9, non-invasive real-time multivariate feature library multivariate application of train vibration source signals
(1) Real-time evaluation of train component service performance
The method comprises the steps of establishing a relation model of vibration signal characteristics in a train room vibration signal characteristic library and train service performance characterization parameters by adopting experimental data, adopting a support vector machine, an extreme learning machine, an artificial neural network, a long-time deep neural network and the like by the model establishing method, outputting and obtaining real-time train service performance characterization parameters by taking the vibration signal characteristics in the train room vibration signal characteristic library obtained based on the step 8 as input of the relation model in the running process of the train, and realizing real-time evaluation of the train component service performance.
(2) Train component failure prediction
And (4) predicting the abnormal vibration condition of the vibration source at a certain moment in the future by adopting a common time series prediction method based on the time series information of the vibration signal characteristics in the train room vibration signal characteristic library established in the step 8, so that the fault prediction of train components at the minute level in the running process of the train can be realized, and the real-time management and control of the safety performance of the train are facilitated.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (10)

1. A train compartment non-invasive vibration monitoring method is characterized by comprising the following steps:
step 1, respectively pre-collecting vibration data M of a plurality of branch measuring points in the train room in the transverse direction, the longitudinal direction and the vertical direction under the condition of normal vibration and abnormal vibration of a known vibration source outside the train room, pre-collecting vibration data C of a total measuring point in the train room in the transverse direction, the longitudinal direction and the vertical direction, and pre-collecting vibration data S of each known vibration source outside the train room in the transverse direction, the longitudinal direction and the vertical direction;
step 2, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data M of each branch measuring point, extracting phase information and amplitude information of first 1-J harmonic signals of the vibration data C of the total measuring point, and extracting phase information and amplitude information of first 1-J harmonic signals of each known vibration source;
step 3, taking the first 1-J harmonic phase difference and the harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the position information of each non-total measuring point relative to the total measuring point in the transverse direction, the longitudinal direction and the vertical direction as output, applying deep echo state network training, adopting the deep echo state network to construct a model for the relationship between the harmonic phase difference and the harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point signal and each non-total measuring point relative to the total measuring point, wherein the model is a model I,
the parameters of the deep echo state network are set as follows, the number of the nodes of the storage pool is set to be 10, 10-fold cross validation is adopted to perform self-adaptive determination on the number of layers of the storage pool and the radius of the matrix spectrum, and the selection range is [1,2,3, …,10 ]]And [0.1,0.2, …,0.9 ]]Selection can accurately reflect the aboveThe parameters of the relationship are used for constructing a nonlinear relationship model f of harmonic phase difference and harmonic amplitude ratio of each non-total measuring point and position information of each non-total measuring point relative to the total measuring point1z(I)，f1h(I)，f1c(I) Respectively representing a longitudinal vibration signal relation model, a transverse vibration signal relation model and a vertical vibration signal relation model;
taking the first 1-J harmonic phase difference between the vibration data of each non-total measuring point and the vibration data of the total measuring point as input, taking the first 1-J harmonic amplitude ratio between the vibration data of each non-total measuring point and the vibration data of the total measuring point as output, training by using a deep echo state network, constructing a model by using the deep echo state network on the relation between the harmonic phase difference and the harmonic amplitude ratio of each non-total measuring point signal relative to the total measuring point signal, wherein the model is a model II,
the parameters of the deep echo state network are set as follows, the number of the nodes of the storage pool is set to be 10, 10-fold cross validation is adopted to carry out self-adaptive determination on the number of layers of the storage pool and the radius of the matrix spectrum, and the selection range is [1,2,3, …,10 ]]And [0.1,0.2, …,0.9 ]]Selecting parameters capable of accurately reflecting the relation, and constructing a relation model f of the harmonic phase shift and the harmonic amplitude ratio of each vibration signal of the non-total measuring points relative to the vibration signal of the total measuring points in the longitudinal direction, the transverse direction and the vertical direction2z(I)，f2h(I)，f2c(I) Respectively representing a longitudinal vibration signal relation model, a transverse vibration signal relation model and a vertical vibration signal relation model,
the non-total measuring point consists of a measuring point and a known vibration source;
step 4, collecting real-time vibration data CR of the total measuring point in the transverse direction, the longitudinal direction and the vertical direction;
and 5, setting an optimization target by taking the phase offset of the vibration source relative to the total measuring point, the combined working state of each known vibration source and the abnormal vibration type as independent variables, executing a multi-objective optimization algorithm based on the first model and the second model, and outputting vibration monitoring result information of the train compartment, wherein the vibration monitoring result information comprises the position of the vibration source generating vibration and/or the abnormal vibration type of the vibration source generating vibration and/or the first 1-J harmonic phase offset of the vibration source generating vibration relative to the total measuring point and/or the first 1-J harmonic amplitude ratio of the vibration source generating vibration relative to the total measuring point.
2. The method for nonintrusive vibration monitoring of train room as claimed in claim 1, wherein in step 1, the normal vibration of the known vibration source outside the train room includes two situations: the first type is that each known vibration source outside the train room normally vibrates, and the second type is that any known vibration source outside the train room independently normally vibrates and the other known vibration sources do not vibrate.
3. The method for nonintrusive vibration monitoring of train room as claimed in claim 1, wherein in step 1, the known vibration source abnormal vibration outside the train room includes two situations: the first type is that any known vibration source outside the train room vibrates independently and the other known vibration sources do not vibrate, and the second type is that any known vibration source outside the train room vibrates independently and the other known vibration sources vibrate normally.
4. The method of nonintrusive vibration monitoring of a railcar room according to claim 1, wherein in said step 1, M ═ LM,m(t)]，C＝[LC,c(t)]，S＝[LS,s(t)](ii) a m (t) is a vibration signal value corresponding to the branch measuring point, c (t) is a vibration signal value corresponding to the total measuring point, and s (t) is a vibration signal value corresponding to the known vibration source; l isMIs a label value L of the vibration data corresponding to the measured pointM＝[lM,lt,lg,ld,lo1]，LCA label value L of the vibration data corresponding to the total measuring pointC＝[ls,lt,lg,ld]，LSTag value L of vibration data corresponding to known vibration sourceS＝[lS,lt,lg,ld,lo2]，lMNumber labels for the distribution points, /)tIs a time stamp tag,/gFor abnormal vibration type label,/o1Is a position information label of the measuring point relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction,/sFor known vibration source tags which vibrate simultaneously,/dFor vibrating direction labels, /)STo be alreadyNumber label of vibration source,/o2The position information labels of the known vibration source relative to the total measuring point in the transverse direction, the longitudinal direction or the vertical direction are obtained.
5. The method for nonintrusive vibration monitoring of train compartments as claimed in claim 4, wherein in step 5, the weighted difference between the sum of the frequency domain characteristics of the vibration signals of the known vibration sources and the frequency domain characteristics of the harmonic orders of the vibration signals of the general measurement point is minimized to be one of the optimization objectives; so as to minimize the weighted variance of the harmonic order differences of the vibration signals of the known vibration sources to be the second optimization objective.
6. The method for nonintrusive vibration monitoring in train compartments as claimed in any one of claims 1 to 5, further comprising step 6 of converting the vibration monitoring result information obtained in step 5 back to a real-time vibration source vibration signal in the time domain.
7. A train compartment vibration signal feature library establishing method, comprising the train compartment non-invasive vibration monitoring method of claim 6, further comprising:
step 7, extracting frequency domain characteristics, time domain characteristics and image characteristics of the real-time vibration source vibration signal, and performing characteristic fusion or characteristic reconstruction on the frequency domain characteristics and the time domain characteristics of the real-time vibration source vibration signal;
and 8, establishing a train compartment vibration signal feature library based on the frequency domain feature, the time domain feature, the image feature, the feature fusion or the feature reconstruction result obtained in the step 7.
8. The method for establishing the train compartment vibration signal feature library of claim 7, further comprising performing on-line training on the model one and/or the model two by using information in the train compartment vibration signal feature library.
9. A method of applying the train compartment vibration signal feature library created by the train compartment vibration signal feature library creating method according to claim 7, comprising:
training to obtain a relation model of vibration signal characteristics in a train compartment vibration signal characteristic library and train service performance characterization parameters, and outputting to obtain real-time train service performance characterization parameters by taking the vibration signal characteristics in the train compartment vibration signal characteristic library obtained in real time as the input of the relation model in the running process of a train.
10. A method of applying the train compartment vibration signal feature library created by the train compartment vibration signal feature library creating method according to claim 7, comprising:
and predicting the abnormal vibration condition of the vibration source at a certain moment in the future by using the time series information of the vibration signal characteristics in the train compartment vibration signal characteristic library and a time series prediction method.
CN202110003658.8A 2021-01-04 2021-01-04 Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method Active CN112816052B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110003658.8A CN112816052B (en) 2021-01-04 2021-01-04 Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110003658.8A CN112816052B (en) 2021-01-04 2021-01-04 Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method

Publications (2)

Publication Number Publication Date
CN112816052A CN112816052A (en) 2021-05-18
CN112816052B true CN112816052B (en) 2022-02-08

Family

ID=75856971

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110003658.8A Active CN112816052B (en) 2021-01-04 2021-01-04 Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method

Country Status (1)

Country Link
CN (1) CN112816052B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113239217B (en) * 2021-06-04 2024-02-06 图灵深视(南京)科技有限公司 Image index library construction method and system, and image retrieval method and system
CN116109786A (en) * 2023-02-16 2023-05-12 中铁四院集团南宁勘察设计院有限公司 Method for constructing urban rail transit vibration map

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103558851A (en) * 2013-10-10 2014-02-05 杨松 Method and device for accurately sensing indoor activities
CN110715783A (en) * 2019-11-04 2020-01-21 合肥工业大学 Method for detecting vibration characteristic of cab of long-head truck
CN111044303B (en) * 2020-01-02 2021-06-04 中车株洲电力机车有限公司 Diagnosis method and device for abnormal vibration of passenger room of maglev train
CN112115643B (en) * 2020-09-15 2022-06-10 中南大学 Smart train service life non-invasive prediction method

Also Published As

Publication number Publication date
CN112816052A (en) 2021-05-18

Similar Documents

Publication Publication Date Title
CN112816052B (en) Train compartment vibration monitoring method, vibration signal characteristic library establishment and application method
Lindemann et al. Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks
CN108053110B (en) PMU data-based transformer state online diagnosis method
CN112800855B (en) Non-invasive real-time fault monitoring method for train bogie
CN110457786B (en) Ship unloader association rule fault prediction model method based on deep confidence network
CN113705412B (en) Track state detection method and device for multi-source data fusion based on deep learning
CN113247041B (en) Wheel set axle box device state monitoring method based on digital twin model
Mathew et al. Regression kernel for prognostics with support vector machines
JP7417342B2 (en) Method for monitoring vibrations in train cabins, construction and application method of vibration signal feature library
CN116380445B (en) Equipment state diagnosis method and related device based on vibration waveform
CN104615121A (en) Method and system for train fault diagnosis
Li et al. Predictive models for influence of primary delays using high‐speed train operation records
Sun et al. Prognostics and health management via long short-term digital twins
Gregori et al. Assessment of catenary condition monitoring by means of pantograph head acceleration and Artificial Neural Networks
KR101242060B1 (en) The Reliability and Availability Management System of the Urban Transit EMU' Maintenance
Dong et al. A quantitative detection method for wheel polygonization of heavy-haul locomotives based on a hybrid deep learning model
Zeng et al. Rail break prediction and cause analysis using imbalanced in-service train data
Wang Unsupervised anomaly detection in railway catenary condition monitoring using autoencoders
CN110610016A (en) Method for predicting rail transit stopping problem based on big data machine learning
CN115222160A (en) Rail transit traction load prediction method based on measured big data
CN115600695A (en) Fault diagnosis method of metering equipment
CN114760194A (en) Fault studying and judging method for Internet of things equipment
Huang et al. Towards automated model calibration and validation in rail transit simulation
Qin et al. Train reliability and safety analysis
Bai et al. Remaining useful life prediction of wheel of heavy-duty railway train based on dual channel multi-scale deep convolution multi-scale deep long short-term memory network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant