CN110222856A - Processing method, device and the storage medium of railway train wheel tread damage - Google Patents
Processing method, device and the storage medium of railway train wheel tread damage Download PDFInfo
- Publication number
- CN110222856A CN110222856A CN201910505741.8A CN201910505741A CN110222856A CN 110222856 A CN110222856 A CN 110222856A CN 201910505741 A CN201910505741 A CN 201910505741A CN 110222856 A CN110222856 A CN 110222856A
- Authority
- CN
- China
- Prior art keywords
- sliding window
- value
- tread
- data
- monitoring data
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Emergency Alarm Devices (AREA)
Abstract
The present invention relates to train overhaul technical fields, disclose processing method, device and the storage medium of a kind of railway train wheel tread damage, solve the problems, such as to can not achieve the early warning of tread damage in the prior art.The described method includes: obtaining the vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel, the monitoring data include Historical Monitoring data and the currently monitored data;Sliding window iteration and smoothing processing are successively carried out to the monitoring data, obtain this corresponding tread damage predicted value of the currently monitored data;Judge whether this described tread damage predicted value is greater than or equal to threshold value of warning;If this described tread damage predicted value is greater than or equal to the threshold value of warning, prompting the selected wheel, there are tread damage failures.The embodiment of the present invention is suitable for the tread damage treatment process of wheel.
Description
Technical field
The present invention relates to train overhaul technical fields, and in particular, to a kind of processing method of railway train wheel tread damage,
Device and storage medium.
Background technique
With the operation of high-speed overload railroad train, wheel tread failure becomes the important event found during vehicle uses
Barrier form, for the monitoring ever more important of tread damage.The main method of existing topical railway wheel problems monitoring includes column inspection people
Work identification and TPDS (Truck Performance Detection System, vehicle riding quality rail side dynamic monitoring system)
Two kinds of alarm, wherein artificial column inspection relies primarily on the experience of train-examiner, omission factor is higher, and manually column inspection can only be found more
Apparent tyre tread failure.TPDS is to comprehensively consider the factors such as wheel weight and speed by the monitoring between impact force wheel track, rushed
Equivalent is hit, so that the railroad train topical railway wheel problems to diversified forms make scientific and reasonable judge.TPDS is examined according to system
The size of the impact equivalent measured realizes alarm to tread damage, but tread damage is often when TPDS alarms
It is more obvious, may have to traffic safety and seriously affect.
Summary of the invention
The purpose of the embodiment of the present invention is that the processing method, device and the storage that provide a kind of railway train wheel tread damage are situated between
Matter solves the problems, such as to can not achieve the early warning of tread damage in the prior art, realizes and utilize TPDS Historical Monitoring number
Early warning is carried out according to tyre tread.
To achieve the goals above, the embodiment of the present invention provides a kind of processing method of railway train wheel tread damage, described
Method includes: to obtain the vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel, the monitoring data
Including Historical Monitoring data and the currently monitored data;Sliding window iteration and smoothing processing are successively carried out to the monitoring data, obtained
This corresponding tread damage predicted value of the currently monitored data;Judge this described tread damage predicted value whether be greater than or
Equal to threshold value of warning;If this described tread damage predicted value is greater than or equal to the threshold value of warning, the selected vehicle is prompted
There are tread damage failures for wheel.
Further, described that sliding window iteration and smoothing processing are successively carried out to the monitoring data, obtain the current prison
This corresponding tread damage predicted value of measured data includes: according to present percussion equivalent and scheduled measurement in the monitoring data
The history impact equivalent of number carries out weight sliding window and adds up, this corresponding sliding window of monitoring data for obtaining this acquisition is cumulative
Value;Index is carried out according to this described sliding window accumulated value and history sliding window accumulated value and assigns power smoothly, obtains the prison of this acquisition
This corresponding tread damage predicted value of measured data.
Further, described to be impacted according to the history of present percussion equivalent and scheduled measurement number in the monitoring data
Equivalent carries out weight sliding window and adds up, this corresponding sliding window accumulated value of monitoring data for obtaining this acquisition includes: according to Wt=
Wc1+n*Wc2, obtain the corresponding the t times sliding window accumulated value of monitoring data of the t times acquisition, wherein WtIt is tired for the t times sliding window
It is value added, Wc1The sum of c1 Secondary Shocks equivalent, W before being the t timesc2The sum of c2 Secondary Shocks equivalent before being the t times, n are weighted value, 0
< n < 1, wherein c1 is less than c2.
Further, described this sliding window accumulated value according to and history sliding window accumulated value carry out index and assign Quan Ping
Sliding, this corresponding tread damage predicted value of monitoring data for obtaining this acquisition includes: basisT >=1 obtains the t times tread damage predicted value Ft+1, wherein WtIt is
T sliding window accumulated value, WiFor the i-th sliding window accumulated value in history sliding window accumulated value, F1It adds up initial value for sliding window, α is smooth
Constant, wherein F1=W1。
Further, this tread damage predicted value described in the judgement whether be greater than or equal to default threshold value of warning it
Afterwards, the method also includes: if this described tread damage predicted value is less than the threshold value of warning, continue to obtain described selected
The monitoring data of wheel.
Further, when the prompt selected wheel is there are after tread damage failure, the method also includes: root
According to speed per hour, load-carrying, impact equivalent and the flat scar depth prediction model of default tyre tread in the currently monitored data, obtain described
The corresponding flat scar depth prediction value of the currently monitored data.
Further, the method also includes: the flat scar depth prediction model of the default tyre tread is determined by following manner:
Obtain the data sample of vehicle corresponding with the selected wheel, the data sample include speed per hour, load-carrying, impact equivalent and
Flat scar depth value;Using the speed per hour, load-carrying, impact equivalent, speed per hour square and the product of equivalent is impacted as BP neural network
Four parameters of input layer, using corresponding flat scar depth value as the output of the BP neural network, training obtains described default
The flat scar depth prediction model of tyre tread.
Correspondingly, the embodiment of the present invention also provides a kind of processing unit of railway train wheel tread damage, described device includes:
Acquiring unit, for obtaining the vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel, the monitoring
Data include Historical Monitoring data and the currently monitored data;Data processing unit, for successively being slided to the monitoring data
Window iteration and smoothing processing obtain this corresponding tread damage predicted value of the currently monitored data;Judging unit, for sentencing
Whether this disconnected described tread damage predicted value is greater than or equal to threshold value of warning;Prompt unit, if being damaged for this described tyre tread
When hurting predicted value more than or equal to the threshold value of warning, prompting the selected wheel, there are tread damage failures.
Further, the data processing unit is also used to according to present percussion equivalent in the monitoring data and makes a reservation for
The history impact equivalent of pendulous frequency carries out weight sliding window and adds up, this corresponding sliding window of monitoring data for obtaining this acquisition is tired
It is value added;Index is carried out according to this described sliding window accumulated value and history sliding window accumulated value and assigns power smoothly, obtains this acquisition
This corresponding tread damage predicted value of monitoring data.
Further, the data processing unit is also used to according to Wt=Wc1+n*Wc2, obtain the monitoring number of the t times acquisition
According to corresponding the t times sliding window accumulated value, wherein WtFor the t times sliding window accumulated value, Wc1C1 Secondary Shocks are worked as before being the t times
The sum of amount, Wc2The sum of c2 Secondary Shocks equivalent before being the t times, n are weighted value, and 0 < n < 1, wherein c1 is less than c2.
Further, the data processing unit is also used to basis
T >=1 obtains the t times tread damage predicted value Ft+1, wherein WtFor the t times sliding window accumulated value, WiFor in history sliding window accumulated value
I-th sliding window accumulated value, F1It adds up initial value for sliding window, α is smoothing constant, wherein F1=W1。
Further, the acquiring unit is also used to judge that this described tread damage predicted value is small when the judging unit
When the threshold value of warning, continue the monitoring data for obtaining the selected wheel.
Further, described device further include: depth prediction unit, for prompting the selected wheel to deposit when prompt unit
After tread damage failure, according to speed per hour, load-carrying, impact equivalent and the flat scar of default tyre tread in the currently monitored data
Depth prediction model obtains the corresponding flat scar depth prediction value of the currently monitored data.
Further, the flat scar depth prediction model of the default tyre tread is determined by following manner: obtains and is selected with described
The data sample of the corresponding vehicle of wheel, the data sample include speed per hour, load-carrying, impact equivalent and flat scar depth value;
Using the speed per hour, load-carrying, impact equivalent, speed per hour square and the product of equivalent is impacted as the input of BP neural network
Four parameters of layer, using corresponding flat scar depth value as the output of the BP neural network, training obtains the default tyre tread
Flat scar depth prediction model.
Correspondingly, the embodiment of the present invention also provides a kind of storage medium, instruction is stored in the storage medium, when its
When being run on computer, so that computer executes the processing method of railway train wheel tread damage as described above.
It adds up through the above technical solutions, the embodiment of the present invention carries out weight sliding window by the TPDS monitoring data to wheel
And index assigns power smoothly, can be realized the early warning of tread damage, and can be using in advance to the tyre tread for reaching threshold value of warning
If the prediction of the flat scar depth of the flat scar depth prediction model realization of tyre tread, obtains flat scar depth prediction value.The embodiment of the present invention solves
The problem of can not achieve the early warning of tread damage in the prior art, realizes the early warning to topical railway wheel problems,
It can quantify the development of tread damage, and not need to increase new monitoring device compared with other algorithms and need to only utilize existing TPDS
Monitoring data.Simultaneously, additionally it is possible to realize the quantitative forecast to wheel flat depth.
Other features and advantages of the present invention will the following detailed description will be given in the detailed implementation section.
Detailed description of the invention
The drawings are intended to provide a further understanding of the invention, and constitutes part of specification, with following tool
Body embodiment is used to explain the present invention together, but is not construed as limiting the invention.In the accompanying drawings:
Fig. 1 is a kind of flow diagram of the processing method of railway train wheel tread damage provided in an embodiment of the present invention;
Fig. 2 is the flow diagram of the processing method of another railway train wheel tread damage provided in an embodiment of the present invention;
Fig. 3 is the exemplary diagram of the result of sliding window accumulated value provided in an embodiment of the present invention;
Fig. 4 is a kind of structure chart of the processing unit of railway train wheel tread damage provided in an embodiment of the present invention;
Fig. 5 is the structure chart of the processing unit of another railway train wheel tread damage provided in an embodiment of the present invention.
Specific embodiment
Below in conjunction with attached drawing, detailed description of the preferred embodiments.It should be understood that this place is retouched
The specific embodiment stated is merely to illustrate and explain the present invention, and is not intended to restrict the invention.
Existing TPDS realizes the alarm to tread damage according to the size of the impact equivalent of the train detected, but works as
Tread damage is often more obvious when TPDS alarms, and is found by the analysis to a large amount of fault cases, very one big
Dividing tread damage failure, there are longer development processes, and the existing alarm method to tread damage does not all account for tyre tread damage
Hurt development process and trend, do not use Historical Monitoring information, can not achieve the early warning to initial failure, and TPDS is not
The quantitative parameter index for providing tread damage of energy.The embodiment of the present invention is exactly the monitoring data using existing TPDS, by right
History impact equivalent data progress weight sliding window is cumulative and index assigns power smoothing processing, then will pass through the entitled sliding window of index
Accumulated value is defined as tread damage predicted value, which can highlight the development of damage as the quantization signifying of tread damage situation
Journey is advantageously implemented early warning.The realization process to the embodiment of the present invention is described below.
Fig. 1 is a kind of flow diagram of the processing method of railway train wheel tread damage provided in an embodiment of the present invention.Such as
Shown in Fig. 1, described method includes following steps:
Step 101, the TPDS monitoring data of selected wheel are obtained, the monitoring data include Historical Monitoring data and current
Monitoring data;
Step 102, sliding window iteration and smoothing processing are successively carried out to the monitoring data, obtains the currently monitored data
This corresponding tread damage predicted value;
Step 103, judge whether this described tread damage predicted value is greater than or equal to threshold value of warning;
Step 104, it when this described tread damage predicted value is greater than or equal to the threshold value of warning, prompts described selected
There are tread damage failures for wheel.
Wherein, the speed per hour in the monitoring data including train, load-carrying and impact equivalent, and the above-mentioned three classes obtained
Data further include the currently monitored data and Historical Monitoring data respectively, that is, include the currently monitored obtained speed per hour, load-carrying and impact
Equivalent further includes speed per hour, load-carrying and the impact equivalent of Historical Monitoring.
When carrying out sliding window iteration and smoothing processing to above-mentioned monitoring data, work as preshoot according in the monitoring data first
The history impact equivalent progress weight sliding window for hitting equivalent and scheduled measurement number is cumulative, obtains the monitoring data pair of this acquisition
This sliding window accumulated value answered.Under normal circumstances, when train wheel tread damage does not occur, impact equivalent that TPDS is monitored
It is 0, starts non-zero impact equivalent occur after there is tread damage, and work as with the impact of the development non-zero of impact injury and measure
Existing frequency gradually rises, and impact equivalent weight values size gradually increases, so the development process reflection of tread damage is cumulative to sliding window
It is exactly that sliding window accumulated value is overall in value constantly to rise, is shown below.
Wt=Wc1+n*Wc2Formula (1)
Using formula (1), the corresponding the t times sliding window accumulated value of monitoring data of available the t times acquisition, wherein Wt
For the t times sliding window accumulated value, Wc1The sum of c1 Secondary Shocks equivalent, W before being the t timesc2C2 Secondary Shocks are worked as before being the t times
The sum of amount, n are weighted value, and 0 < n < 1, wherein c1 is less than c2.Wherein, Wc1Size can preferably reflect the current state of wheel,
But for untreated tread damage Wc2The historical failure situation of wheel can be preferably reacted, this is to Historical Monitoring information
Primary coarse tax power, and reflect monitoring data situation in the period of the t times preceding length of window with sliding window accumulated value.Example
Such as, with t for certain sliding window accumulated value, for c1 5, c2 20, n are 0.5, formula (1) is represented by W=W5+0.5*W20, i.e.,
W5The sum of the impact equivalent of 5 measurements before indicating this t times, W20The impact of 20 measurements before indicating this t times is worked as
The sum of amount.For the setting of above-mentioned c1, c2 and n can be according to the distribution density for being directed to train difference line sniffing station, vehicles
Frequency of use be arranged, in the case of or detection density high for frequency of use is big, length of window should be lengthened suitably, be
The big route of detection density may take c1=40, c2=10.In addition, in the case of extreme, when in first time, measurement impact is worked as
When obtaining first time sliding window accumulated value after amount, zero that number is c2 is filled before the impact equivalent of this measurement, with
Just use when the sum of c2 Secondary Shocks equivalent before calculating the t times.
Since the frequency of occurrences of tread damage initial stage non-zero impact equivalent is lower and more discrete, cause to impact equivalent
Sliding window accumulated value WtFluctuation it is larger, therefore to obtained sliding window accumulated value carry out index assign power smoothing processing and carry out a step
Prediction is realized and is weighed along time dimension to the secondary tax of Historical Monitoring data.
Wherein, it is as follows to assign power formula for index:
Ft+1=α Wt+(1-α)FtFormula (2)
Wherein, Ft+1For the t times tread damage predicted value, FtFor the t-1 times tread damage predicted value, WtFor the t times sliding window
Accumulated value, α are smoothing constant, and 0 < α < 1, can be derived by following formula by formula (2):
Ft=α Wt-1+(1-α)Ft-1Formula (3)
Formula (3) are substituted into formula (2) and are obtained:
Ft+1=α Wt+(1-α)[αWt-1+(1-α)Ft-1] formula (4)
And the t-2 times tread damage predicted value are as follows:
Ft-1=α Wt-2+(1-α)Ft-2Formula (5)
Formula (5) are substituted into formula (4), are obtained:
Ft+1=α Wt+(1-α)αWt-1+(1-α)2αWt-2+(1-α)3Ft-2Formula (6)
Wherein, F1=W1, t >=1, F1It adds up initial value for sliding window, then available general term formula:
In actual application, it since sequence length may be very long, calculates, the sum of first few items can be made in order to simplify
For tread damage predicted value, ignore weight below close to 0 item.
As can be seen from the above formula that the t times tread damage predicted value Ft+1With the t-1 times tread damage predicted value FtHave
It closes, influence of the distance the t times remoter sliding window accumulated value to the t times tread damage predicted value is smaller.Not only by smoothing processing
Reduce the fluctuation of sliding window accumulated value, tread damage predicted value can also be obtained, there is directive significance for tread damage early warning.
The prediction of this tread damage is defined as this sliding window accumulated value is assigned the smooth value obtained later of power by index
Value, for reflecting the development process and severity of tread damage.Threshold value of warning by the way that tread damage is arranged is realized to stepping on
The early warning of surface damage.Judge whether this described tread damage predicted value is greater than or equal to threshold value of warning, if this described tyre tread
When damage forecast value is less than the threshold value of warning, then continues to execute step 101 and obtain the monitoring data of the selected wheel, and hold
The following process flow of row.If this described tread damage predicted value is greater than or equal to the threshold value of warning, then institute is prompted
Stating selected wheel, there are tread damage failures.Wherein, by analyzing a large amount of fault cases, the threshold value of warning of tread damage is arranged
More in time tread damage failure can be forecast between 200-250.
In the driving process of train, wheel flat is that flat sliding, removing, defect, slag etc. influence wheel rolling circle
Degree, causes the general designation of wheel periodicity bump rail failure.With the rotation of wheel, the wheel of flat scar occurs in train driving mistake
Intermittent pulse excitation source is generated to wheel track in journey, when wheel is rolled at flat scar, the impact force of generation is several times greater than usual,
Sometimes even more than ten times.This will deteriorate dynamics of vehicle performance, aggravate the fair wear and tear and rail wear of bogie part,
Bearing service life is influenced, increases vehicle and runs noise intensity, gently then influence vehicle smoothness, increases wheel rotation and repairs work
Amount, it is heavy then cause hot axis, fervent, seriously affect traffic safety.Therefore, when this described tread damage predicted value is greater than or equal to
It, can be according to speed per hour, load-carrying, impact equivalent and the flat scar of default tyre tread in the currently monitored data after the threshold value of warning
Depth prediction model obtains the corresponding flat scar depth prediction value of the currently monitored data.
Wherein, when determining the flat scar depth prediction model of the default tyre tread, the BP neural network of pre-training can be passed through
Flat scar depth prediction is carried out, the severity of the quantitative evaluation tread damage of flat scar depth is utilized.BP neural network be it is a kind of by
It is current most widely used neural network according to the multilayer feedforward neural network of error backpropagation algorithm training.BP nerve net
Network has arbitrarily complicated pattern classification ability and excellent multidimensional function mapping ability, and solving simple perceptron not can solve
The problem of.
In embodiments of the present invention, BP neural network is for the data sample of training during vehicle practice
To the scene feedback measurement data of TPDS alarm, it is also possible to measured data of experiment.Data sample vehicle should be flat with actual prediction
The vehicle of scar depth prediction value is same vehicle.Wherein, the data sample includes speed per hour when vehicle passes through, load-carrying, impact
The flat scar depth value that equivalent and in-site measurement obtain.By the correlation analysis of feature, by speed per hour, load in TPDS monitoring data
Weight, impact equivalent are denoted as V (km/h), T (t), P respectively, take V, T, P, V2* tetra- features of P are as characteristic parameter, flat scar depth value
For fitting amount, training sample set size is not less than 200.
Since the physical significance that different characteristic indicates is different, there is also larger differences for the order of magnitude, it is therefore desirable to data into
Row normalized, standardized method are standard deviation scaling common in the art, can effectively reduce outlier
It influences.
The flat scar depth prediction model of the default tyre tread is built using BP neural network, network structure is input layer 4 sections
Point, 4 every layer of hidden layers, 40 node, one node of output layer use relu function as activation letter in the 2nd, 4 hidden layers
Number, is defined as mean square error, learning rate 0.1 for loss function.After building neural network according to above-mentioned setting, carry out 8000 times
Training obtains the flat scar depth prediction model of the default tyre tread.
It, can be by the currently monitored number after this described tread damage predicted value is greater than or equal to the threshold value of warning
Speed per hour, load-carrying in, impact equivalent, speed per hour square and the product of equivalent is impacted as the default flat scar depth prediction of tyre tread
The input of model, the default flat scar depth prediction model output of tyre tread is flat scar depth prediction value.In addition, the present invention is implemented
The prediction error of the default flat scar depth prediction model of tyre tread in example by verifying is in 0.15mm or so.
The embodiment of the present invention is cumulative by the TPDS monitoring data progress weight sliding window to wheel and index assigns power smoothly,
It can be realized the early warning of tread damage, and can be using the default flat scar depth prediction of tyre tread to the tyre tread for reaching threshold value of warning
The prediction of the flat scar depth of model realization, obtains flat scar depth prediction value.The embodiment of the present invention solves in the prior art cannot be real
The problem of early warning of existing tread damage, the early warning to topical railway wheel problems is realized, tread damage can be quantified
Development, and do not need to increase new monitoring device compared with other algorithms and need to only utilize existing TPDS monitoring data.Simultaneously, moreover it is possible to
Enough quantitative forecasts realized to wheel flat depth.
Embodiment to facilitate the understanding of the present invention, Fig. 2 are a kind of railway train wheel tread damages provided in an embodiment of the present invention
Processing method flow chart.As shown in Fig. 2, described method includes following steps:
Step 201, the vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel, the prison are obtained
Measured data includes Historical Monitoring data and the currently monitored data;
Step 202, equivalent is impacted according to the history of present percussion equivalent and scheduled measurement number in the monitoring data
It carries out weight sliding window to add up, obtains this corresponding sliding window accumulated value of monitoring data of this acquisition;
Step 203, index is carried out according to this described sliding window accumulated value and history sliding window accumulated value and assigns power smoothly, obtained
This this corresponding tread damage predicted value of monitoring data obtained;
Step 204, judge whether this described tread damage predicted value is greater than or equal to threshold value of warning, be to then follow the steps
205, otherwise return step 201;
Step 205, prompting the selected wheel, there are tread damage failures;
Step 206, according to speed per hour, load-carrying, impact equivalent and the flat scar depth of default tyre tread in the currently monitored data
Prediction model is spent, the corresponding flat scar depth prediction value of the currently monitored data is obtained.
By taking the wheel of the 7th wheel of certain C80 type open-top car as an example, since the wheel needs to carry out to need to carry out because of shelled tread
Face and repairs, then when finding that non-zero impact equivalent occurs for the first time in the wheel in the TPDS monitoring data for searching the wheel, impact
Equivalent is 17.According to formula Wt=W5+0.5*W20, as shown in figure 3, the wheel is obtained after there is non-zero impact equivalent, it is whole
The result of the sliding window accumulated value detected every time during a fault progression.
As can be seen from the figure gradually development trend is presented in tread damage, and the fluctuation of failure early stage sliding window accumulated value is more obvious
It is unfavorable for early warning.Therefore index tax power is carried out to sliding window accumulated value and smoothly obtains this tread damage predicted value.This is calculated
Obtained sliding window accumulated value is updated in formula (7), and wherein α value is 0.2.After assigning power smoothly by index, preferable
Reservation original data while, reduce data fluctuations.In addition, by the data cases of analyzing a large amount of fault cases by early warning threshold
Value is set as 200, when this tread damage predicted value reaches 200 or is greater than 200, then suggests emphasis inspection, while being stepped on
The flat scar depth prediction of surface damage, in conjunction with the severity of depth prediction result evaluation tread damage.
Wherein, when building the flat scar depth prediction model of default tyre tread using BP neural network, network structure is input layer 4
A node, 4 every layer of hidden layers, 40 node, one node of output layer use Relu function as activation in the 2nd, 4 hidden layers
Loss function is defined as mean square error, learning rate 0.1 by function.After building BP neural network, carry out 8000 times it is trained
To the flat scar depth prediction model of default tyre tread, input speed per hour in the currently monitored data, load-carrying, impact equivalent, speed per hour square with
The product for impacting equivalent, obtains the corresponding flat scar depth prediction value of the currently monitored data.
It, not only can benefit when determining the flat scar depth prediction model of default tyre tread in addition, expansible, in the embodiment of the present invention
With BP neural network, Rbf nerve, svm regression algorithm etc. can also be utilized.
Correspondingly, Fig. 4 is a kind of structure chart of the processing unit of railway train wheel tread damage provided in an embodiment of the present invention.
As shown in figure 4, described device 40 includes acquiring unit 41, for obtaining the vehicle riding quality rail side dynamic monitoring of selected wheel
System TPDS monitoring data, the monitoring data include Historical Monitoring data and the currently monitored data;Data processing unit 42 is used
In successively carrying out sliding window iteration and smoothing processing to the monitoring data, this corresponding tyre tread of the currently monitored data is obtained
Damage forecast value;Judging unit 43, for judging whether this described tread damage predicted value is greater than or equal to threshold value of warning;It mentions
Show unit 44, if be greater than or equal to the threshold value of warning for this described tread damage predicted value, prompts the selected vehicle
There are tread damage failures for wheel.
Wherein, the data processing unit is also used to according to present percussion equivalent and scheduled measurement in the monitoring data
The history impact equivalent of number carries out weight sliding window and adds up, this corresponding sliding window of monitoring data for obtaining this acquisition is cumulative
Value;Index is carried out according to this described sliding window accumulated value and history sliding window accumulated value and assigns power smoothly, obtains the prison of this acquisition
This corresponding tread damage predicted value of measured data.
In addition, the data processing unit is also used to according to Wt=Wc1+n*Wc2, obtain the monitoring data pair of the t times acquisition
The t times sliding window accumulated value answered, wherein WtFor the t times sliding window accumulated value, Wc1Be the t times before c1 Secondary Shocks equivalent it
With Wc2The sum of c2 Secondary Shocks equivalent before being the t times, n are weighted value, and 0 < n < 1, wherein c1 is less than c2.
In addition, the data processing unit is also used to basist≥
1, obtain the t times tread damage predicted value Ft+1, wherein WtFor the t times sliding window accumulated value, WiFor in history sliding window accumulated value
I sliding window accumulated value, F1It adds up initial value for sliding window, α is smoothing constant, wherein F1=W1。
Wherein, the acquiring unit is also used to judge that this described tread damage predicted value is less than institute when the judging unit
When stating threshold value of warning, continue the monitoring data for obtaining the selected wheel.
As shown in figure 5, described device further include: depth prediction unit 45, for prompting the selected vehicle when prompt unit
Wheel is there are after tread damage failure, according to speed per hour, load-carrying, impact equivalent and the default tyre tread in the currently monitored data
Flat scar depth prediction model obtains the corresponding flat scar depth prediction value of the currently monitored data.
Wherein, the flat scar depth prediction model of the default tyre tread can be determined by following manner: being obtained and the selected vehicle
The data sample of corresponding vehicle is taken turns, the data sample includes speed per hour, load-carrying, impact equivalent and flat scar depth value;By institute
The product of speed per hour, load-carrying, impact equivalent, speed per hour square and impact equivalent is stated as BP neural network four of input layer join
Number, using corresponding flat scar depth value as the output of the BP neural network, it is pre- that training obtains the flat scar depth of the default tyre tread
Survey model.
The embodiment of the present invention is cumulative by the TPDS monitoring data progress weight sliding window to wheel and index assigns power smoothly,
It can be realized the early warning of tread damage, and can be using the default flat scar depth prediction of tyre tread to the tyre tread for reaching threshold value of warning
The prediction of the flat scar depth of model realization, obtains flat scar depth prediction value.The embodiment of the present invention solves in the prior art cannot be real
The problem of early warning of existing tread damage, the early warning to topical railway wheel problems is realized, tread damage can be quantified
Development, and do not need to increase new monitoring device compared with other algorithms and need to only utilize existing TPDS monitoring data.Simultaneously, moreover it is possible to
Enough quantitative forecasts realized to wheel flat depth.
Present apparatus operating process, referring to the realization process of the processing method of above-mentioned railway train wheel tread damage.
Correspondingly, the embodiment of the present invention also provides a kind of storage medium, instruction is stored in the storage medium, when its
When being run on computer, so that computer executes the processing method of the damage of railway train wheel tread described in above-described embodiment.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/
Or the forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable Jie
The example of matter.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including element
There is also other identical elements in process, method, commodity or equipment.
It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The above is only embodiments herein, are not intended to limit this application.To those skilled in the art,
Various changes and changes are possible in this application.It is all within the spirit and principles of the present application made by any modification, equivalent replacement,
Improve etc., it should be included within the scope of the claims of this application.
Claims (15)
1. a kind of processing method of railway train wheel tread damage, which is characterized in that the described method includes:
The vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel are obtained, the monitoring data include going through
History monitoring data and the currently monitored data;
Sliding window iteration and smoothing processing are successively carried out to the monitoring data, corresponding this of the currently monitored data is obtained and steps on
Surface damage predicted value;
Judge whether this described tread damage predicted value is greater than or equal to threshold value of warning;
If this described tread damage predicted value is greater than or equal to the threshold value of warning, prompting the selected wheel, there are tyre treads
Damage fault.
2. the method according to claim 1, wherein it is described the monitoring data are successively carried out sliding window iteration and
Smoothing processing, obtaining this corresponding tread damage predicted value of the currently monitored data includes:
Weight sliding window is carried out according to the history of present percussion equivalent and scheduled measurement number impact equivalent in the monitoring data
It is cumulative, obtain this corresponding sliding window accumulated value of monitoring data of this acquisition;
Index is carried out according to this described sliding window accumulated value and history sliding window accumulated value and assigns power smoothly, obtains the prison of this acquisition
This corresponding tread damage predicted value of measured data.
3. according to the method described in claim 2, it is characterized in that, it is described according to present percussion equivalent in the monitoring data with
And the history impact equivalent of scheduled measurement number carries out that weight sliding window is cumulative, obtain this acquisition monitoring data it is corresponding this
Sliding window accumulated value includes:
According to Wt=Wc1+n*Wc2, obtain the corresponding the t times sliding window accumulated value of monitoring data of the t times acquisition, wherein WtFor institute
State the t times sliding window accumulated value, Wc1The sum of c1 Secondary Shocks equivalent, W before being the t timesc2Be the t times before c2 Secondary Shocks equivalent it
It is weighted value with, n, 0 < n < 1, wherein c1 is less than c2.
4. according to the method described in claim 2, it is characterized in that, described this sliding window accumulated value according to and history are sliding
Window accumulated value carries out index and assigns power smoothly, this corresponding tread damage predicted value of monitoring data for obtaining this acquisition includes:
According toObtain the t times tread damage predicted value Ft+1,
In, WtFor the t times sliding window accumulated value, WiFor the i-th sliding window accumulated value in history sliding window accumulated value, F1It is cumulative initial for sliding window
Value, α is smoothing constant, wherein F1=W1。
5. the method according to claim 1, wherein this tread damage predicted value described in the judgement whether
After default threshold value of warning, the method also includes:
If this described tread damage predicted value is less than the threshold value of warning, continue the monitoring number for obtaining the selected wheel
According to.
6. the method according to claim 1, wherein when there are tread damage events for the prompt selected wheel
After barrier, the method also includes:
According to speed per hour, load-carrying, impact equivalent and the flat scar depth prediction model of default tyre tread in the currently monitored data, obtain
To the corresponding flat scar depth prediction value of the currently monitored data.
7. according to the method described in claim 6, it is characterized in that, the method also includes:
The flat scar depth prediction model of the default tyre tread is determined by following manner:
The data sample of vehicle corresponding with the selected wheel is obtained, the data sample includes speed per hour, load-carrying, impacts equivalent
And flat scar depth value;
Using the speed per hour, load-carrying, impact equivalent, speed per hour square and the product of equivalent is impacted as the input layer of BP neural network
Four parameters, using corresponding flat scar depth value as the output of the BP neural network, training obtains the flat scar of default tyre tread
Depth prediction model.
8. a kind of processing unit of railway train wheel tread damage, which is characterized in that described device includes:
Acquiring unit, it is described for obtaining the vehicle riding quality rail side dynamic monitoring system TPDS monitoring data of selected wheel
Monitoring data include Historical Monitoring data and the currently monitored data;
Data processing unit obtains the current prison for successively carrying out sliding window iteration and smoothing processing to the monitoring data
This corresponding tread damage predicted value of measured data;
Judging unit, for judging whether this described tread damage predicted value is greater than or equal to threshold value of warning;
If prompt unit prompts the choosing be greater than or equal to the threshold value of warning for this described tread damage predicted value
Determining wheel, there are tread damage failures.
9. device according to claim 8, which is characterized in that the data processing unit is also used to according to the monitoring number
Weight sliding window is carried out according to the history of middle present percussion equivalent and scheduled measurement number impact equivalent to add up, and obtains this acquisition
This corresponding sliding window accumulated value of monitoring data;Index is carried out according to this described sliding window accumulated value and history sliding window accumulated value
It assigns power smoothly, obtains this corresponding tread damage predicted value of monitoring data of this acquisition.
10. device according to claim 9, which is characterized in that the data processing unit is also used to according to Wt=Wc1+n*
Wc2, obtain the corresponding the t times sliding window accumulated value of monitoring data of the t times acquisition, wherein WtFor the t times sliding window accumulated value,
Wc1The sum of c1 Secondary Shocks equivalent, W before being the t timesc2The sum of c2 Secondary Shocks equivalent before being the t times, n are weighted value, 0 < n < 1,
Wherein c1 is less than c2.
11. device according to claim 9, which is characterized in that the data processing unit is also used to basisObtain the t times tread damage predicted value Ft+1, wherein WtIt is
T sliding window accumulated value, WiFor the i-th sliding window accumulated value in history sliding window accumulated value, F1It adds up initial value for sliding window, α is smooth
Constant, wherein F1=W1。
12. device according to claim 8, which is characterized in that the acquiring unit is also used to sentence when the judging unit
When this disconnected described tread damage predicted value is less than the threshold value of warning, continue the monitoring data for obtaining the selected wheel.
13. device according to claim 8, which is characterized in that described device further include:
Depth prediction unit, for prompting the selected wheel there are after tread damage failure, according to described when prompt unit
Speed per hour, load-carrying, impact equivalent and the flat scar depth prediction model of default tyre tread in the currently monitored data, obtain the current prison
The corresponding flat scar depth prediction value of measured data.
14. device according to claim 13, which is characterized in that
The flat scar depth prediction model of the default tyre tread is determined by following manner:
The data sample of vehicle corresponding with the selected wheel is obtained, the data sample includes speed per hour, load-carrying, impacts equivalent
And flat scar depth value;
Using the speed per hour, load-carrying, impact equivalent, speed per hour square and the product of equivalent is impacted as the input layer of BP neural network
Four parameters, using corresponding flat scar depth value as the output of the BP neural network, training obtains the flat scar of default tyre tread
Depth prediction model.
15. a kind of storage medium, which is characterized in that instruction is stored in the storage medium, when run on a computer,
So that computer executes the processing method of the described in any item railway train wheel tread damages of the claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910505741.8A CN110222856B (en) | 2019-06-12 | 2019-06-12 | Method and device for processing damage to train wheel tread and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910505741.8A CN110222856B (en) | 2019-06-12 | 2019-06-12 | Method and device for processing damage to train wheel tread and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110222856A true CN110222856A (en) | 2019-09-10 |
CN110222856B CN110222856B (en) | 2022-03-25 |
Family
ID=67816706
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910505741.8A Active CN110222856B (en) | 2019-06-12 | 2019-06-12 | Method and device for processing damage to train wheel tread and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110222856B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111023998A (en) * | 2019-12-11 | 2020-04-17 | 神华铁路货车运输有限责任公司 | Tread damage measuring device, method, apparatus and storage medium |
CN114056381A (en) * | 2021-11-24 | 2022-02-18 | 西南交通大学 | Railway vehicle wheel flat scar monitoring method |
CN114789743A (en) * | 2022-06-22 | 2022-07-26 | 成都铁安科技有限责任公司 | Method and system for monitoring abnormal operation of train wheels |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102060036A (en) * | 2010-12-14 | 2011-05-18 | 成都主导科技有限责任公司 | System for detecting rapid train wheel pair dynamically |
US20140200951A1 (en) * | 2013-01-11 | 2014-07-17 | International Business Machines Corporation | Scalable rule logicalization for asset health prediction |
CN104506804A (en) * | 2014-12-22 | 2015-04-08 | 安徽三联交通应用技术股份有限公司 | Device and method for monitoring abnormal behavior of motor vehicle on expressway |
CN107133478A (en) * | 2017-05-10 | 2017-09-05 | 南京航空航天大学 | A kind of high speed incremental formula aero-engine method for detecting abnormality |
-
2019
- 2019-06-12 CN CN201910505741.8A patent/CN110222856B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102060036A (en) * | 2010-12-14 | 2011-05-18 | 成都主导科技有限责任公司 | System for detecting rapid train wheel pair dynamically |
US20140200951A1 (en) * | 2013-01-11 | 2014-07-17 | International Business Machines Corporation | Scalable rule logicalization for asset health prediction |
CN104506804A (en) * | 2014-12-22 | 2015-04-08 | 安徽三联交通应用技术股份有限公司 | Device and method for monitoring abnormal behavior of motor vehicle on expressway |
CN107133478A (en) * | 2017-05-10 | 2017-09-05 | 南京航空航天大学 | A kind of high speed incremental formula aero-engine method for detecting abnormality |
Non-Patent Citations (1)
Title |
---|
刘杰宇: "列车轮对踏面故障监控系统研究", 《中国优秀硕士论文全文数据库》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111023998A (en) * | 2019-12-11 | 2020-04-17 | 神华铁路货车运输有限责任公司 | Tread damage measuring device, method, apparatus and storage medium |
CN114056381A (en) * | 2021-11-24 | 2022-02-18 | 西南交通大学 | Railway vehicle wheel flat scar monitoring method |
CN114789743A (en) * | 2022-06-22 | 2022-07-26 | 成都铁安科技有限责任公司 | Method and system for monitoring abnormal operation of train wheels |
CN114789743B (en) * | 2022-06-22 | 2022-09-16 | 成都铁安科技有限责任公司 | Method and system for monitoring abnormal running of train wheels |
Also Published As
Publication number | Publication date |
---|---|
CN110222856B (en) | 2022-03-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Mohammadi et al. | Exploring the impact of foot-by-foot track geometry on the occurrence of rail defects | |
CN110222856A (en) | Processing method, device and the storage medium of railway train wheel tread damage | |
KR102069087B1 (en) | Apparatus and method for predicting train accident risk using fault probability | |
Li et al. | Estimation of railway track longitudinal irregularity using vehicle response with information compression and Bayesian deep learning | |
Ye et al. | Shock detection of rotating machinery based on activated time-domain images and deep learning: An application to railway wheel flat detection | |
Si et al. | Fault prediction model based on evidential reasoning approach | |
Appoh et al. | Hybrid adaptive model to optimise components replacement strategy: A case study of railway brake blocks failure analysis | |
Shubinsky et al. | Application of machine learning methods for predicting hazardous failures of railway track assets | |
CN104933488A (en) | Fault rate prediction method of ATP (Automatic Train Protection) vehicle-mounted equipment | |
Li et al. | A data-driven approach to RUL prediction of tools | |
Schafer | Effect of train length on railroad accidents and a quantitative analysis of factors affecting broken rails | |
Peng et al. | A coordinated ramp metering framework based on heterogeneous causal inference | |
JP5987749B2 (en) | Component failure occurrence prediction method, component failure occurrence prediction program, and component failure occurrence prediction device | |
Lee et al. | Generative adversarial network-based missing data handling and remaining useful life estimation for smart train control and monitoring systems | |
Zhang et al. | Broken rail prediction with machine learning-based approach | |
Zeng et al. | Rail break prediction and cause analysis using imbalanced in-service train data | |
Vagnoli et al. | An automatic bridge damage diagnostics method using empirical mode decomposition based health indicators and neuro‐fuzzy classification | |
US20220343701A1 (en) | Terrain classification and system health monitoring | |
CN116611864A (en) | Urban rail transit train speed and passenger flow prediction method based on RNN | |
Awad et al. | Predicting urban rail transit safety via artificial neural networks | |
Niu et al. | A hybrid bearing prognostic method with fault diagnosis and model fusion | |
Lasisi et al. | Rail track quality and T-stochastic neighbor embedding for hybrid track index | |
Zhang et al. | Railway train wheel maintenance model and its application | |
Mal et al. | Modern condition monitoring systems for railway wheel-set dynamics: Performance analysis and limitations of existing techniques | |
Zhang et al. | Forecast method of track irregularity of heavy-haul railway based on bp neural 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 |