CN113511183A - Optimization criterion-based early fault separation method for air brake system of high-speed train - Google Patents

Optimization criterion-based early fault separation method for air brake system of high-speed train Download PDF

Info

Publication number
CN113511183A
CN113511183A CN202110798141.2A CN202110798141A CN113511183A CN 113511183 A CN113511183 A CN 113511183A CN 202110798141 A CN202110798141 A CN 202110798141A CN 113511183 A CN113511183 A CN 113511183A
Authority
CN
China
Prior art keywords
fault
speed train
sample
index
brake system
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.)
Granted
Application number
CN202110798141.2A
Other languages
Chinese (zh)
Other versions
CN113511183B (en
Inventor
纪洪泉
王建东
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.)
Shandong University of Science and Technology
Original Assignee
Shandong University of Science and Technology
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 Shandong University of Science and Technology filed Critical Shandong University of Science and Technology
Priority to CN202110798141.2A priority Critical patent/CN113511183B/en
Publication of CN113511183A publication Critical patent/CN113511183A/en
Application granted granted Critical
Publication of CN113511183B publication Critical patent/CN113511183B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60T—VEHICLE BRAKE CONTROL SYSTEMS OR PARTS THEREOF; BRAKE CONTROL SYSTEMS OR PARTS THEREOF, IN GENERAL; ARRANGEMENT OF BRAKING ELEMENTS ON VEHICLES IN GENERAL; PORTABLE DEVICES FOR PREVENTING UNWANTED MOVEMENT OF VEHICLES; VEHICLE MODIFICATIONS TO FACILITATE COOLING OF BRAKES
    • B60T17/00—Component parts, details, or accessories of power brake systems not covered by groups B60T8/00, B60T13/00 or B60T15/00, or presenting other characteristic features
    • B60T17/18—Safety devices; Monitoring
    • B60T17/22—Devices for monitoring or checking brake systems; Signal devices
    • B60T17/228—Devices for monitoring or checking brake systems; Signal devices for railway vehicles
    • 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

  • Engineering & Computer Science (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Valves And Accessory Devices For Braking Systems (AREA)

Abstract

The invention discloses an optimization criterion-based early fault separation method for an air brake system of a high-speed train, which belongs to the field of fault diagnosis and comprises the following steps: collecting pressure measurement data of a plurality of groups of brake cylinders under the normal working condition of a high-speed train to form a plurality of training data sets; selecting a brake holding stage, calculating a fault detection combination index of a corresponding sample contained in each training data set, and calculating a threshold value; under the real-time working condition of the high-speed train, acquiring brake cylinder pressure data meeting requirements as a test sample, calculating a fault detection combination index of the test sample, and comparing the fault detection combination index with a threshold value to judge whether an abnormality exists or not; if the fault occurs, reconstructing the fault sample in different fault directions based on an optimization criterion, and if the combined index is lower than a threshold value after reconstruction, determining that a fault source is found, so as to realize fault separation. The invention is a data driving method without accurate modeling of the system, has universality for the air brake system of the high-speed train with the same or similar mechanism, and can simultaneously realize the functions of detection and separation.

Description

Optimization criterion-based early fault separation method for air brake system of high-speed train
Technical Field
The invention belongs to the field of fault diagnosis, and particularly relates to an optimization criterion-based early fault separation method for an air brake system of a high-speed train.
Background
In recent years, the high-speed rail industry in China is rapidly developed, and the achievement of drawing attention is achieved. By the end of 2020, the operating mileage of high-speed rail in China is nearly 4 kilometers and stably stays at the first position of the world. As the core composition of a high-speed rail system, a high-speed train is closely related to the personal safety of passengers, and the safe and reliable operation of the high-speed train is very important. The braking system of the high-speed train mainly realizes that the train can complete the specified deceleration and stop under normal and emergency conditions, and is an extremely important safety key system. The braking mode of the high-speed train mainly comprises two modes of electric braking and air braking: electric braking has the advantage of energy recovery, but air brake technology remains very important and indispensable as a final insurance.
At present, a single-variable overrun alarm mechanism, namely KNORR logic adopted by the current actual operation vehicle, is mainly adopted for pressure monitoring of the air brake system of the high-speed train. Still other self-test circuits are capable of diagnosing pressure sensor hardware faults such as short circuits and open circuits. The monitoring mode can ensure the stable operation of the train air brake system by matching with a fault guiding safety mechanism adopted by the train, but is not sensitive to several common early faults (especially when the fault amplitude is small) in the air brake system, and cannot effectively monitor the early faults. Early failure diagnosis has the significance, both can kill the trouble in the sprouting state, avoids further evolution to serious trouble, can also provide the fault diagnosis result for the engineer in order to realize the accurate replacement and the maintenance of condition of part, uses manpower sparingly, material resources.
CN110293949A discloses a method for detecting minor faults of an air brake system of a high-speed train, which provides a minor fault detection method based on mixed indexes aiming at several common minor faults in the air brake system of the high-speed train, and explains the effectiveness of strategies through experimental results. However, the monitoring strategy disclosed in the above patent only implements a fault detection function, is limited in technical difficulty and does not provide a fault separation (i.e. fault location) method, that is, cannot automatically find a component in which an abnormality occurs.
In summary, a new early fault separation method is needed to perform online monitoring and diagnosis on the air brake system, and accurately complete the positioning of several types of early faults in the air brake system.
Disclosure of Invention
In order to solve the problems, the invention provides an optimization criterion-based early fault separation method for an air brake system of a high-speed train, which is used for carrying out online monitoring and diagnosis on the air brake system, accurately positioning the abnormity to a component and realizing the positioning of early faults.
The technical scheme of the invention is as follows:
the method for separating the early fault of the air brake system of the high-speed train based on the optimization criterion comprises the following steps:
acquiring a plurality of groups of brake cylinder pressure measurement data including a normal braking process under the historical operating condition of a high-speed train to form a plurality of training data sets;
step two, selecting a stable deceleration stage of the braking process, calculating a fault detection combination index of each corresponding sample contained in each training data set in the step one, and solving a threshold value of the fault detection combination index;
selecting a stable deceleration stage of the braking process under the actual operation condition of the high-speed train, collecting the pressure data of the brake cylinder at the current moment in real time as a test sample, calculating a fault detection combination index of the sample, and comparing the fault detection combination index with the threshold value in the step two to judge whether an early fault occurs;
and step four, if the early fault is judged to occur in the step three, reconstructing the current abnormal sample in different fault directions based on the optimization criterion, and if the reconstructed fault detection combination index is lower than the threshold value in the step two, judging that the fault source is found, so that early fault separation is realized.
Preferably, the specific process of the step one is as follows:
on the basis of historical operation monitoring data of the high-speed train, under the condition that an air brake system of the high-speed train normally works, collecting pressure measurement data of a brake cylinder of each train in the braking process to form a plurality of training data sets; suppose that q training data sets are collected in total, and are respectively marked as a matrix X1，X2，…，XqEach line of the training data set represents the measured values of the pressure of a plurality of brake cylinders at the corresponding sampling moment, and the number of the brake cylinders is set to be m; the number of rows in each matrix represents the number of samples contained in the data set, assumed to be Np,p＝1,2,...,q。
Preferably, the specific process of step two is as follows:
only paying attention to and screening out samples in each data set at a stable deceleration stage in the braking process based on the data set in the step one; any satisfactory training sample is recorded as x ═ x1,x2,...,xm]TWherein x isiRepresenting the pressure value of the ith brake cylinder of a certain sample in the brake maintaining stage;
and calculating a fault detection combination index by adopting the following formula:
Figure BDA0003163549010000021
wherein x issThe desired set point, representing pressure during the brake hold phase, is a known quantity;
Figure BDA0003163549010000022
the average value of the m brake cylinder pressures contained in the sample is represented, and the average value represents the central values of different brake cylinder pressures; the first term (x) in the failure detection combination index calculation formula (1)i-xs)2Indicating the degree of deviation of the brake cylinder pressure from the setpoint, second term
Figure BDA0003163549010000023
The dispersion degree of the values of the different brake cylinder pressures is represented;
substituting all samples meeting the requirements in the q training data sets in the step one into a formula (1) to obtain a combined index value of each sample, and assuming that the total number of the samples meeting the requirements is N, then a threshold eta of the combined index is2The following formula is obtained:
Figure BDA0003163549010000031
the threshold is set as the maximum of all N sample fault detection combined indicators.
Preferably, the specific process of step three is as follows:
Gao Suwhen the train actually runs and the braking process reaches the braking maintaining stage, the pressure data of the brake cylinder at the current moment are sequentially collected as test samples and recorded as xtEach element represents the pressure of the corresponding brake cylinder, and the meaning of each element is consistent with that of the element in the training sample in the step two; according to the definition formula (1) of the fault detection combination index in the step two, the test sample x is solvedtThe combined index is taken, and whether an early fault occurs is judged according to a threshold value; if Index (x)t)＞η2If not, the air brake system of the high-speed train is judged to be in a normal running state.
Preferably, the specific process of step four is as follows:
according to the judgment result of the third step, under the premise of fault occurrence, reconstructing the combination index of the samples along the possible fault directions based on the optimization criterion, and assuming that J possible fault directions are total, and is marked as { xi-jJ ═ 1,2,. ·, J }; xi along any jth direction-jOptimizing and calculating to obtain a reconstruction index value
Figure BDA0003163549010000032
Figure BDA0003163549010000033
Wherein I represents an identity matrix with dimension m; m represents a square matrix of dimension M: the values of diagonal elements are (2m-1)/m, and the values of off-diagonal elements are-1/m; xijRepresenting the fault direction, the number of rows is m, and the number of columns depends on the specific fault mode; vector xsDimension is m, all element values are the ideal pressure set value x in the step twos；
Pairs Index (x) along all the failure directions mentioned abovet)＞η2Optimizing to obtain J optimized values
Figure BDA0003163549010000034
If an optimized value is lower than the threshold eta2If the fault reconstruction is correct, the direction is determinedThe direction in which the fault actually occurs is specified, thereby achieving early fault separation.
The invention has the following beneficial technical effects:
the invention provides an early fault separation strategy, which is a data driving method based on the operation principle of an air brake system of a high-speed train. The method only needs to utilize historical and online pressure monitoring data, and does not need a complex system mechanism model, thereby being convenient for practical application; because the adopted combined index comprises two parts, the pressure tracking performance and the discrete degrees of different pressure values are respectively measured, the strategy can monitor multiple types of early faults in the air brake system, such as single sensor faults, multiple sensor faults and relay valve faults; based on the optimization criterion, the provided fault separation method can position the abnormality to the element part, thereby providing valuable information for the overhaul and maintenance of the air brake system of the high-speed train.
Drawings
FIG. 1 is a flow chart of a method for early fault separation of an air brake system of a high speed train based on optimization criteria;
FIG. 2 is a schematic view of the detection result of the early failure of the single sensor of the air brake system in embodiment 1;
FIG. 3 is a schematic diagram showing the result of early failure separation of a single sensor of the air brake system in embodiment 1;
FIG. 4 is a schematic view of the detection result of the early failure of the multiple sensors of the air brake system in embodiment 1;
fig. 5 is a schematic diagram of the result of early failure separation of multiple sensors of the air brake system in embodiment 1.
Detailed Description
The invention is described in further detail below with reference to the following figures and detailed description:
the invention aims to realize the automatic separation of the early fault of the air brake system of the high-speed train and provide valuable information for the maintenance and the repair of the air brake system according to the situation, thereby saving the time and the resources for manual overhaul and test. The preamble of fault separation is fault detection. The fault detection statistic adopted by the invention is a combined index containing two meanings, so that the index can have a good detection effect on various common early faults. The fault separation index is obtained based on a strict optimization theory, so that the fault can be positioned and traced.
As shown in FIG. 1, the method for separating the early fault of the air brake system of the high-speed train based on the optimization criterion comprises the following steps:
step S110: collecting a plurality of groups of brake cylinder pressure measurement data including a normal braking process under the historical operating condition of the high-speed train to form a plurality of training data sets;
specifically, in the actual running process of the high-speed train, the data acquisition and network transmission system can automatically record and store some key measurement information; under the condition that an expert judges that no abnormality occurs in an air brake system of the high-speed train, collecting historical brake cylinder pressure measurement data generated during the operation of the high-speed train to form a plurality of training data sets; usually, the high-speed train comprises a plurality of trains, and the training data sets can be derived from multiple operations of one train or multiple operations of the plurality of trains; suppose that q sets of training data are collected together and are respectively recorded as matrix X1，X2，…，XqEach line of the training data set represents the measured values of the pressure of the brake cylinders at the corresponding sampling moment, and the number of the brake cylinders is set to be m; the number of rows in each matrix represents the number of samples contained in the data set, assumed to be Np,p＝1,2,...,q。
Step S120: selecting a stable deceleration stage of the braking process, calculating a fault detection combination index of a corresponding sample contained in each training data set in the step S110, and solving a threshold value of the combination index;
specifically, based on the data sets in step S110, only samples in each data set that are in the smooth deceleration stage (i.e., the brake hold stage) of the braking process are focused and screened out; each group in the q groups of training data sets has a section of data meeting the requirements, and the sections of the data meeting the requirements in different sets are different in length; without loss of generality, any of the above will be usedThe satisfactory training samples are denoted x ═ x1,x2,...,xm]TWherein x isiThe pressure value of the ith brake cylinder of a certain sample in the brake maintaining stage is represented; the following fault detection combination indexes are adopted:
Figure BDA0003163549010000041
wherein x issIndicating the desired set value of the pressure during the brake hold phase, which set value differs at different brake levels, but according to expert experience xsIs a known amount;
Figure BDA0003163549010000051
and the average value of m brake cylinder pressures contained in the sample represents the central values of different brake cylinder pressures.
The first term (x) in the above-mentioned fault detection combination index (1)i-xs)2Indicating the degree of deviation of the brake cylinder pressure from the setpoint, second term
Figure BDA0003163549010000052
And the dispersion degree of the values of the different brake cylinder pressures is represented.
Substituting all samples meeting the requirements in the q training data sets in the step S110 into a formula (1) to obtain a combined index value of each sample, and assuming that the total number of the samples meeting the requirements is N, then a threshold eta of the combined index is2The following formula is obtained:
Figure BDA0003163549010000053
that is, the threshold is set to the maximum value among all the N sample combination indexes.
Step S130: under the actual operation condition of the high-speed train, selecting a stable deceleration stage in the braking process, collecting the pressure data of the brake cylinder at the current moment in real time as a test sample, calculating a fault detection combination index of the sample, and comparing the fault detection combination index with the threshold value in the step S120 to judge whether an early fault occurs;
specifically, when the actual operation of the high-speed train is carried out, when the braking process reaches a braking maintaining stage, the pressure data of the brake cylinder at the current moment are sequentially collected to be used as test samples, and the test samples are recorded as xtEach element represents the pressure of the corresponding brake cylinder, and the meaning of each element is consistent with that of the element in the training sample in the step S120; according to the definition (1) of the fault detection combination index in step S120, the test sample x is obtainedtThe combined index value is taken, and whether an early fault occurs is judged according to a threshold value; specifically, if Index (x)t)＞η2If not, the air brake system of the high-speed train is judged to be in a normal running state.
Step S140: if the step S130 determines that an early fault occurs, reconstructing the current abnormal sample in different fault directions based on an optimization criterion, and if the reconstructed combined index is lower than the threshold of the step S120, determining that a fault source is found, thereby realizing early fault separation;
specifically, according to the judgment result of step S130, on the premise that a failure occurs, i.e., Index (x)t)＞η2(ii) a Reconstructing the combined index of the samples along the possible failure directions based on the optimization criterion, assuming a total of J possible failure directions, denoted { XIjJ ═ 1,2,. ·, J }; the possible fault directions can be obtained according to experience of a first-line engineer expert, and for a single sensor fault type, there are m possible fault directions; xi along any jth direction-jOptimizing and calculating to obtain the following reconstruction index value
Figure BDA0003163549010000054
Figure BDA0003163549010000055
The formula (3) is obtained based on a strict solution optimization problem; wherein I represents an identity matrix with dimension m; m is a square matrix with dimension M: the diagonal elements take the values of (2m-1)The values of off-diagonal elements are-1/m; xijRepresenting the fault direction, the number of rows is m, the number of columns depends on the specific fault form: if only a single sensor failure occurs, the number of columns is 1, and if a double independent sensor failure occurs, the number of columns is 2; vector xsDimension is m, and all the values of the elements are the ideal pressure set value x in the step S120s；
Pairs Index (x) along all the failure directions mentioned abovet)＞η2Optimizing to obtain J optimized values
Figure BDA0003163549010000061
If an optimized value is lower than the threshold eta2Then the fault reconstruction is considered to be correct, and the direction is designated as the actual fault direction, so that early fault separation is realized.
The method is essentially a data-driven monitoring method on the basis of effectively utilizing the structure and the operation mechanism of the air brake system of the high-speed train, so that the complex and accurate modeling of the components of the air brake system is not needed, and the method is easy to be practically applied. On the other hand, most of the air brake systems of high-speed trains have almost the same or similar structures and operation mechanisms, so that the method has considerable universality.
Compared with the prior art, the method can simultaneously realize the detection and the separation of several types of early faults of the air brake system of the high-speed train, thereby not only indicating whether the system has the abnormity, but also providing important information for the abnormity occurrence position.
Example 1
To further understand the present invention and to show the visual effect of the method of the present invention for early failure separation of air brake system of high speed train, a detailed description is given below with a specific embodiment. The experiment related to the embodiment is based on a high-speed train braking system joint debugging test bed of a central Qingdao tetragonal research institute company, Inc. It is worth to be noted that the method provided by the invention is not only suitable for the test bed, but also suitable for other high-speed train air brake systems with similar working mechanisms. The specific process is as follows:
(1) historical data set collection
In practice, data generated by the operation of the high-speed train is recorded and stored. Aiming at the brake test bed (consistent with an actual train brake system) related to the embodiment, the whole brake process can be simulated for many times under the condition that the system has no fault, and corresponding measurement data is recorded by utilizing monitoring software matched with the test bed. And selecting the first train of the test bed without loss of generality, selecting the highest level under the conventional braking at the braking level, and collecting multiple groups of brake cylinder pressure measurement data. The platform train contains 4 simulated brake cylinders and is equipped with corresponding measuring sensors, so in this example m is 4.
(2) Calculating threshold value of fault detection combination index and reserving for use
Based on the multiple groups of training data sets collected in the step (1), firstly, screening out sample sections in the brake holding stage in each group of data sets according to expert experience and measured data characteristics. Then, all the screened samples are substituted into the formula (1), and the combination index thereof is calculated. In this example, the pressure set point is x for the highest braking level under conventional brakings299. Then, the combined index threshold is calculated by formula (2) to obtain η2＝13.75。
(3) Collecting real-time monitoring data, and judging whether early failure occurs
The real-time monitoring data is brake cylinder pressure measurement data obtained by the measurement of a sensor carried by a brake system in the actual running process of the high-speed train. In the test bed, the braking process can be artificially simulated, and the pressure data of the brake cylinder can be obtained by using a sensor arranged on the test bed. During the simulation, faults, such as valve or pipeline faults, can be applied in advance, and additive fault types for the open-loop sensor can also be obtained by the subsequent addition of measured data. And (3) inputting a brake cylinder pressure measurement sample in a brake maintaining stage in the braking process into a formula (1) to calculate a fault detection index, comparing the fault detection index with the threshold value obtained in the step (2), and if the index value of the real-time monitoring sample exceeds 13.75, determining that an early fault occurs.
(4) Segregating early failures
When the sample is judged to be abnormal in step (3), all possible failure directions are determined first and are marked as { xijJ ═ 1,2,. said, J }. For example, for a single sensor fault type, because there are 4 sensors in total, there are 4 possible fault directions, xijRespectively, all column vectors of the 4-dimensional identity matrix. For these possible failure directions, reconstruction index values in a certain direction are respectively calculated according to equation (3). If an optimized value is lower than the threshold eta2If the fault reconstruction is deemed correct 13.75, the direction is designated as the actual fault direction, and early fault separation is achieved.
In this embodiment, based on the brake system joint debugging test bed, we explore the detection and separation of two types of early faults, namely single sensor fault and double sensor fault. FIGS. 2 to 5 are schematic views showing specific detection results and separation results, and the abscissa of FIGS. 2 to 5 indicates sample xtThe ordinate of fig. 2 and 4 is the value of the combination index, the ordinate of fig. 3 and 5 is the value of the combination index after optimization, and the broken lines of fig. 2 to 5 are all the set threshold values η2。
First, the detection and separation effect of early failure of a single sensor is explored. And setting the sensor fault on the No. 1 sensor, wherein the fault type is a deviation fault and the amplitude is 4 kPa. A braking process was run for a total of 70 sampling instants, where the fault occurred at the 31 st sample. Fig. 2 shows the detection effect of the combined index on the fault, and it can be seen that the fault can be detected in time after the fault occurs. The four subgraphs in fig. 3 show the reconstructed index values of all 70 test samples after optimization along four sensor directions, where R is1Indicating along the first sensor xi1＝[1000]TAnd reconstructing the direction, and so on. As can be seen from FIG. 3, only R is present after the fault has occurred1Index R below threshold and reconstructed along other three directions2、 R3、R4Still greater than the threshold value, so sensor number 1 is determined as the faulty sensor, as opposed to the actual situationAnd (4) sign.
Secondly, the detection and separation effect of the early failure of the multiple sensors is explored. And the sensor faults are respectively and independently set on the No. 2 sensor and the No. 3 sensor, the fault types are still deviation faults, and the amplitude values are all 3 kPa. A braking process was run once for a total of 80 sampling instants, where the fault occurred at the 21 st sample. Fig. 4 shows the detection result of the fault by the combined index, and it can be seen that the multiple sensor fault can also be detected in time. With respect to the optimization of the combined index of abnormal samples, the reconstructed index cannot be made smaller than the threshold value along 4 single sensor directions. Next, a double sensor failure type is tried, 2 sensor failures are arbitrarily selected out of 4 sensors, and there are a total of
Figure RE-GDA0003258781870000071
Xi Zi respectively1＝[1 1 0 0]T、 Ξ2＝[1 0 1 0]T、Ξ3＝[1 0 0 1]T、Ξ4＝[0 1 1 0]T、Ξ5＝[0 1 0 1]T、Ξ6＝[0 0 1 1]T. The six subgraphs in FIG. 5 show the reconstructed index values of all the test samples after being optimized along the six directions, where R is1Denotes along a first direction xi1＝[1 1 0 0]TAnd reconstructing the direction, and so on. As can be seen from FIG. 5, only R is present after the fault has occurred4Lower than the threshold value, while the index reconstructed along the other direction is still larger than the threshold value, so xi4The corresponding No. 2 and No. 3 sensors are determined as sensors with faults and are consistent with the actual situation. Thus, the effectiveness of the method of the present invention is demonstrated.
It is to be understood that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make modifications, alterations, additions or substitutions within the spirit and scope of the present invention.

Claims (5)

1. The method for separating the early fault of the air brake system of the high-speed train based on the optimization criterion is characterized by comprising the following steps of:
acquiring a plurality of groups of brake cylinder pressure measurement data including a normal braking process under the historical operating condition of a high-speed train to form a plurality of training data sets;
step two, selecting a stable deceleration stage of the braking process, calculating a fault detection combination index of each corresponding sample contained in each training data set in the step one, and solving a threshold value of the fault detection combination index;
selecting a stable deceleration stage of the braking process under the actual operation condition of the high-speed train, collecting the pressure data of the brake cylinder at the current moment in real time as a test sample, calculating a fault detection combination index of the sample, and comparing the fault detection combination index with the threshold value in the step two to judge whether an early fault occurs;
and step four, if the early fault is judged to occur in the step three, reconstructing the current abnormal sample in different fault directions based on the optimization criterion, and if the reconstructed fault detection combination index is lower than the threshold value in the step two, judging that the fault source is found, so that early fault separation is realized.
2. The optimization criterion-based early fault separation method for the air brake system of the high-speed train is characterized in that the specific process of the step one is as follows:
on the basis of historical operation monitoring data of the high-speed train, under the condition that an air brake system of the high-speed train normally works, collecting pressure measurement data of a brake cylinder of each train in the braking process to form a plurality of training data sets; suppose that q training data sets are collected in total and are respectively recorded as matrix X1，X2，…，XqEach line of the training data set represents the measured values of the pressure of a plurality of brake cylinders at the corresponding sampling moment, and the number of the brake cylinders is set to be m; the number of rows in each matrix represents the number of samples contained in the data set, assumed to be Np,p＝1,2,...,q。
3. The optimization criterion-based early fault separation method for the air brake system of the high-speed train according to claim 2, wherein the specific process of the second step is as follows:
only paying attention to and screening out samples in each data set at a stable deceleration stage in the braking process based on the data set in the step one; any satisfactory training sample is recorded as x ═ x1,x2,...,xm]TWherein x isiThe pressure value of the ith brake cylinder of a certain sample in the brake maintaining stage is represented;
and calculating a fault detection combination index by adopting the following formula:
Figure FDA0003163547000000011
wherein x issThe desired set point, representing pressure during the brake hold phase, is a known quantity;
Figure FDA0003163547000000012
the average value of the m brake cylinder pressures contained in the sample is represented, and the average value represents the central values of different brake cylinder pressures; the first term (x) in the fault detection combination index calculation formula (1)i-xs)2Indicating the degree of deviation of the brake cylinder pressure from the setpoint, second term
Figure FDA0003163547000000013
The dispersion degree of the pressure values of different brake cylinders is represented;
substituting all samples meeting the requirements in the q training data sets in the step one into a formula (1) to obtain a combined index value of each sample, and assuming that the total number of the samples meeting the requirements is N, then a threshold eta of the combined index is2The following formula is obtained:
Figure FDA0003163547000000021
the threshold is set as the maximum of all N sample fault detection combined indicators.
4. The optimization criterion-based early fault separation method for the air brake system of the high-speed train is characterized in that the specific process of the step three is as follows:
when the high-speed train actually runs and the braking process reaches the braking maintaining stage, the pressure data of the brake cylinder at the current moment are sequentially collected to be used as a test sample and recorded as xtEach element represents the pressure of the corresponding brake cylinder, and the meaning of each element is consistent with that of the element in the training sample in the step two; according to the definition formula (1) of the fault detection combination index in the step two, the test sample x is solvedtThe combined index value is taken, and whether an early fault occurs is judged according to a threshold value; if Index (x)t)＞η2If not, the air brake system of the high-speed train is judged to be in a normal running state.
5. The optimization criterion-based early fault separation method for the air brake system of the high-speed train is characterized in that the specific process of the step four is as follows:
according to the judgment result of the third step, under the premise of fault occurrence, reconstructing the combination index of the samples along the possible fault directions based on the optimization criterion, and assuming that J possible fault directions are total, and is marked as { xi-jJ ═ 1,2,. ·, J }; xi along any jth direction-jOptimizing and calculating to obtain a reconstruction index value
Figure FDA0003163547000000022
Figure FDA0003163547000000023
Wherein I represents an identity matrix with dimension m; m represents a square matrix of dimension M: the values of diagonal elements are (2m-1)/m, and the values of off-diagonal elements are-1/m; xijRepresenting the fault direction, the number of rows is m, and the number of columns depends on the specific fault mode; vector xsThe dimension of the number of the dimension m,all the element values are the ideal pressure set value x in the step twos；
Pairs Index (x) along all the failure directions mentioned abovet)＞η2Optimizing to obtain J optimized values
Figure FDA0003163547000000024
If an optimized value is lower than the threshold eta2Then the fault reconstruction is considered to be correct, and the direction is designated as the actual fault direction, so that early fault separation is realized.
CN202110798141.2A 2021-07-15 2021-07-15 Optimization criterion-based early fault separation method for air brake system of high-speed train Active CN113511183B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110798141.2A CN113511183B (en) 2021-07-15 2021-07-15 Optimization criterion-based early fault separation method for air brake system of high-speed train

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110798141.2A CN113511183B (en) 2021-07-15 2021-07-15 Optimization criterion-based early fault separation method for air brake system of high-speed train

Publications (2)

Publication Number Publication Date
CN113511183A true CN113511183A (en) 2021-10-19
CN113511183B CN113511183B (en) 2022-05-17

Family

ID=78067106

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110798141.2A Active CN113511183B (en) 2021-07-15 2021-07-15 Optimization criterion-based early fault separation method for air brake system of high-speed train

Country Status (1)

Country Link
CN (1) CN113511183B (en)

Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05233463A (en) * 1991-11-15 1993-09-10 Fujitsu Ltd Method and device for displaying fault of storage device array
KR20100093668A (en) * 2009-02-17 2010-08-26 주식회사 만도 Method for storing diagnosis data in electric brake system
CN103853144A (en) * 2012-11-29 2014-06-11 沈阳工业大学 On-site sensor fault detection method based on oil-extraction production data
CN104035431A (en) * 2014-05-22 2014-09-10 清华大学 Obtaining method and system for kernel function parameters applied to nonlinear process monitoring
CN104111454A (en) * 2014-07-09 2014-10-22 电子科技大学 Scanning radar angular super-resolution imaging method
US20150067012A1 (en) * 2013-08-28 2015-03-05 Infineon Technologies Ag Method and data processing device for reconstructing a vector
CN104842990A (en) * 2014-10-24 2015-08-19 北汽福田汽车股份有限公司 Vehicle brake detection system and method
CN105259895A (en) * 2015-10-14 2016-01-20 山东科技大学 Method and monitoring system for detecting and separating micro fault in industrial process
CN105730431A (en) * 2016-01-29 2016-07-06 清华大学 Failure monitoring method and failure monitoring system for brake cylinders of motor train unit
CN108092623A (en) * 2017-12-16 2018-05-29 西安理工大学 A kind of photovoltaic array multisensor fault detecting and positioning method
CN109672406A (en) * 2018-12-20 2019-04-23 福州大学 A method of photovoltaic power generation array fault diagnosis and classification based on rarefaction representation and SVM
CN109871000A (en) * 2019-02-25 2019-06-11 山东科技大学 A kind of closed loop industrial process sensor method for diagnosing faults of data-driven
JP2019090749A (en) * 2017-11-16 2019-06-13 富士通株式会社 Arrival direction estimating device and arrival direction estimating method
CN110033135A (en) * 2019-04-15 2019-07-19 北京交通大学 The train braking system failure prediction method that Multivariate Time Series feature is reinforced
CN110293949A (en) * 2019-06-06 2019-10-01 山东科技大学 A kind of bullet train Pneumatic brake systems small fault detection method
CN111324105A (en) * 2020-02-27 2020-06-23 山东科技大学 Small sensor fault detection and separation method based on statistic Mahalanobis distance

Patent Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05233463A (en) * 1991-11-15 1993-09-10 Fujitsu Ltd Method and device for displaying fault of storage device array
KR20100093668A (en) * 2009-02-17 2010-08-26 주식회사 만도 Method for storing diagnosis data in electric brake system
CN103853144A (en) * 2012-11-29 2014-06-11 沈阳工业大学 On-site sensor fault detection method based on oil-extraction production data
US20150067012A1 (en) * 2013-08-28 2015-03-05 Infineon Technologies Ag Method and data processing device for reconstructing a vector
CN104035431A (en) * 2014-05-22 2014-09-10 清华大学 Obtaining method and system for kernel function parameters applied to nonlinear process monitoring
CN104111454A (en) * 2014-07-09 2014-10-22 电子科技大学 Scanning radar angular super-resolution imaging method
CN104842990A (en) * 2014-10-24 2015-08-19 北汽福田汽车股份有限公司 Vehicle brake detection system and method
CN105259895A (en) * 2015-10-14 2016-01-20 山东科技大学 Method and monitoring system for detecting and separating micro fault in industrial process
CN105730431A (en) * 2016-01-29 2016-07-06 清华大学 Failure monitoring method and failure monitoring system for brake cylinders of motor train unit
JP2019090749A (en) * 2017-11-16 2019-06-13 富士通株式会社 Arrival direction estimating device and arrival direction estimating method
CN108092623A (en) * 2017-12-16 2018-05-29 西安理工大学 A kind of photovoltaic array multisensor fault detecting and positioning method
CN109672406A (en) * 2018-12-20 2019-04-23 福州大学 A method of photovoltaic power generation array fault diagnosis and classification based on rarefaction representation and SVM
CN109871000A (en) * 2019-02-25 2019-06-11 山东科技大学 A kind of closed loop industrial process sensor method for diagnosing faults of data-driven
CN110033135A (en) * 2019-04-15 2019-07-19 北京交通大学 The train braking system failure prediction method that Multivariate Time Series feature is reinforced
CN110293949A (en) * 2019-06-06 2019-10-01 山东科技大学 A kind of bullet train Pneumatic brake systems small fault detection method
CN111324105A (en) * 2020-02-27 2020-06-23 山东科技大学 Small sensor fault detection and separation method based on statistic Mahalanobis distance

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
刘佳音: "SVD降噪与排列熵方法在齿轮故障诊断中的应用", 《中国优秀硕士学位论文全文数据库（电子期刊）工程科技II辑》 *
周东华等: "Fault Detection and Isolation of the Brake Cylinder System for Electric Multipe Unites", 《 IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY》 *
周东华等: "高速列车信息控制***的故障诊断技术", 《自动化学报》 *
滕薇: "矿压监测***中传感器故障检测与分离算法研究", 《中国优秀硕士学位论文全文数据库（电子期刊）工程科技I辑》 *
郑彦涛: "基于层次分析法的CRH380B动车组维修可靠性分析与研究", 《中国优秀硕士学位论文全文数据库（电子期刊）工程科技II辑》 *

Also Published As

Publication number Publication date
CN113511183B (en) 2022-05-17

Similar Documents

Publication Publication Date Title
CN110293949B (en) Method for detecting tiny faults of air brake system of high-speed train
CN108569607B (en) Elevator fault early warning method based on bidirectional gating cyclic neural network
EP0391047B1 (en) Brake-testing and -monitoring procedure in railway cars, and arrangement for executing the procedure
CN111114519B (en) Railway vehicle brake fault prediction method and health management system
CN104134010B (en) Satellite fault diagnosis method for discrete type data based on Naive Bayes
CN109358598A (en) Pure electric automobile off-line test device and method based on UDS diagnosing protocol
CN105730431B (en) EMU checking cylinder fault monitoring method and fault monitoring system
CN102797671A (en) Fault detection method and device of reciprocating compressor
CN108092623B (en) A kind of photovoltaic array multisensor fault detecting and positioning method
CN105204496A (en) Sensor fault diagnosing method and system for air braking control system of motor train unit
DE4438854C2 (en) Monitoring system for a technical system
CN111071291B (en) Train wheel set monitoring system and train wheel set monitoring method
CN109507992B (en) Method, device and equipment for predicting faults of locomotive brake system components
CN116047207A (en) Nuclear power plant electric valve performance fault diagnosis and assessment method
CN103853144B (en) Spot sensor fault detection method based on oil recovery creation data
CN110554682B (en) Fault detection reasoning method based on fault correlation analysis
CN113511183B (en) Optimization criterion-based early fault separation method for air brake system of high-speed train
CN110097219A (en) A kind of electric vehicle O&M optimization method based on security tree model
CN110084500A (en) Motor vehicle security control method and electric vehicle based on safety tree probability and security-critical degree
CN109556861A (en) A kind of bearing real-time fault diagnosis system of case-based reasioning
CN112944603A (en) Air conditioner health assessment method for rail transit
DE202021100535U1 (en) Fault diagnosis and maintenance device for a braking system of a high-speed train
CN112198462A (en) Traction transformer abnormal state identification method based on railway dispatching operation plan
CN110879152A (en) Liquid rocket engine test real-time data diagnosis strategy
Chen et al. EPBS_FIDMV: A fault injection and diagnosis methods validation benchmark for EPBS of EMU

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