CN109918687A - A kind of train dynamics emulation mode and emulation platform based on machine learning - Google Patents
A kind of train dynamics emulation mode and emulation platform based on machine learning Download PDFInfo
- Publication number
- CN109918687A CN109918687A CN201711296331.4A CN201711296331A CN109918687A CN 109918687 A CN109918687 A CN 109918687A CN 201711296331 A CN201711296331 A CN 201711296331A CN 109918687 A CN109918687 A CN 109918687A
- Authority
- CN
- China
- Prior art keywords
- data information
- attribute data
- default label
- label
- train
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the present invention provides a kind of train dynamics emulation mode and emulation platform based on machine learning.Wherein, this method comprises: according to default tag extraction attribute data information relevant to the default label；The default label is demarcated according to the attribute data information；Using the attribute data information and the default label as training set, is exercised supervision study based on support vector regression algorithm, obtain and the regression curve of the default label is obtained by the attribute data information；Using the regression curve, the value of label is preset according to the attribute data information prediction, realizes the dynamics simulation simulation of virtual train.The embodiment of the present invention is trained the attribute data information and label of train by the support vector regression algorithm in machine learning, realizes simulation and test to virtual train on emulation platform；Time and the manpower consumption of the monitoring of train field monitoring can be reduced, and improves efficiency and accuracy rate in entire test process.
Description
Technical field
The present embodiments relate to computer simulation technique field, specially a kind of train dynamics based on machine learning
Emulation mode and emulation platform.
Background technique
Machine learning is a kind of using computer simulation human behavior, to obtain the subject of new knowledge and skills, is used
Machine learning can automatically find the value wherein contained from data, and can convert people for primary data information (pdi) can use
Information.In rail traffic, exploitation set of system applies to train operation, and primary work is exactly train real-time testing system
Stability and reliability, general test are that scene operation debugging, data board record train operating data and state to train many times,
In this case, it requires a great deal of time and real-time monitoring monitoring is carried out to train with manpower.
In urban track traffic, the prior art is using genetic algorithm to train operation simulation.When using genetic algorithm,
The inaccuracy that the lack of standardization of its coding may be faced and indicated, and single genetic algorithm encoding cannot comprehensively will be excellent
The constraint expression of change problem comes out.On the other hand, the efficiency of genetic algorithm is relatively low, easy Premature Convergence, to the precision of algorithm, can
Row degree, computational complexity etc., without effective quantitative analysis method.
Summary of the invention
To solve the problems, such as train simulation in the prior art, the embodiment of the present invention provides a kind of based on machine learning
Train dynamics emulation mode and emulation platform.
In a first aspect, the embodiment of the present invention provides a kind of train dynamics emulation mode based on machine learning, this method
It include: according to default tag extraction attribute data information, the attribute data information is related to the default label；According to described
Attribute data information demarcates the default label；Using the attribute data information and the default label as training set, base
It exercises supervision study in support vector regression algorithm, it is bent to obtain the recurrence that the default label is obtained by the attribute data information
Line；Using the regression curve, the value of label is preset according to the attribute data information prediction, realizes the dynamic of virtual train
Mechanics Simulation simulation.
Second aspect, the embodiment of the present invention provide a kind of train dynamics emulation platform based on machine learning, the emulation
Platform includes: attribute data data obtaining module, is specifically used for according to default tag extraction attribute data information, the attribute number
It is believed that breath is related to the default label；Default label demarcating module is specifically used for demarcating institute according to the attribute data information
State default label；Machine learning module is specifically used for using the attribute data information and the default label as training set,
It is exercised supervision study based on support vector regression algorithm, obtains and the recurrence of the default label is obtained by the attribute data information
Curve；Simulation and prediction module is specifically used for utilizing the regression curve, be marked in advance according to the attribute data information prediction
The value of label realizes the dynamics simulation simulation of virtual train.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, including memory and processor, the processor and
The memory completes mutual communication by bus；The memory, which is stored with, to be referred to by the program that the processor executes
It enables, the processor calls described program instruction to be able to carry out following method: according to default tag extraction attribute data information, institute
It is related to the default label to state attribute data information；The default label is demarcated according to the attribute data information；Using institute
Attribute data information and the default label are stated as training set, is exercised supervision study, obtained based on support vector regression algorithm
The regression curve of the default label is obtained by the attribute data information；Using the regression curve, according to the attribute number
According to the value for presetting label described in information prediction, the dynamics simulation simulation of virtual train is realized.
Fourth aspect, the embodiment of the present invention provide a kind of computer readable storage medium, are stored thereon with computer program,
The computer program realizes following method when being executed by processor: according to default tag extraction attribute data information, the attribute
Data information is related to the default label；The default label is demarcated according to the attribute data information；Utilize the attribute
Data information and the default label are exercised supervision study based on support vector regression algorithm, are obtained by described as training set
Attribute data information obtains the regression curve of the default label；Using the regression curve, according to the attribute data information
The value for predicting the default label realizes the dynamics simulation simulation of virtual train.
The embodiment of the present invention is by the support vector regression algorithm in machine learning, to the attribute data information and mark of train
Label are trained, and realize simulation and test to virtual train on emulation platform；Can reduce train field monitoring monitoring when
Between and manpower consumption, and improve efficiency and accuracy rate in entire test process.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the train dynamics emulation mode flow chart provided in an embodiment of the present invention based on machine learning；
Fig. 2 is the train dynamics simulation platform structure schematic diagram provided in an embodiment of the present invention based on machine learning；
Fig. 3 is the structural schematic diagram of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
Fig. 1 is the train dynamics emulation mode flow chart provided in an embodiment of the present invention based on machine learning.Such as Fig. 1 institute
Show, which comprises
Step 101, basis preset tag extraction attribute data information, the attribute data information and the default label phase
It closes；
The default label refers to by emulating the parameter to be predicted.According to the default label from the data board number of train
It is described related to the default label to refer to the attribute number according to middle extraction attribute data information relevant to the default label
It is believed that the value of breath has an impact to the value of the default label.
The train real time information parameter of train head end and tail end vehicle-mounted recording when the data board data are train operation；It can
The data board data are stored entirely in local, so that simulated program calls the attribute data information needed；It can also be direct
Extract the attribute data information that simulated program needs from the data board data, such as described category relevant to the default label
Property data information.After extracting the attribute data information, by the attribute data information preservation of extraction in local or database
For simulated program calling.
Step 102 demarcates the default label according to the attribute data information；
The attribute data information is on the influential data of the value of default label tool, according to the kinetic simulation of train
Type or empirical equation obtain the relationship of the attribute data information Yu the default label, and then according to the attribute data information
The default label is demarcated, namely the default label is calculated according to the attribute data information.
Step 103, using the attribute data information and the default label as training set, be based on support vector regression
Algorithm exercises supervision study, obtains and obtains the regression curve of the default label by the attribute data information；
Support vector regression algorithm (SVM) is a kind of sample learning algorithm for having solid theoretical basis, it avoids complexity
Deductive procedure, realize efficiently from training sample to the inference mode of forecast sample, SVR, which is simplified, common to be returned, is pre-
Survey problem reduces the complexity of mass data processing, substantially increases training effectiveness, is better than genetic algorithm.
It is input with the attribute data information and the default label when the attribute data information is various dimensions,
It is output with the default label, is trained using support vector regression algorithm；Wherein, the attribute data as input
Information and the default label as output obtain the regression curve for training；The default label as input
For supervision machine learning training process, during fitting obtains the regression curve, if according to the regression curve
The gap of the predicted value of the obtained default label and the value of the actual default label exceeds preset threshold, then gives up phase
Answer sample.
When being analyzed using support vector regression, according to the magnanimity history number of the attribute data information and the default label
According to being trained and being fitted, obtains and the regression curve of the default label is obtained by the attribute data information.
Particularly, if in any period the attribute data information or default label, when the attribute number it is believed that
It is input with temporal information when breath or default label are single dimension, is defeated with the attribute data information or the default label
Out, it is trained using support vector regression, the available attribute data information or the default label change over time
Model.
Step 104, using the regression curve, the value of label is preset according to the attribute data information prediction, it is real
The dynamics simulation simulation of existing virtual train；
After acquisition obtains the regression curve of the default label by the attribute data information, for the new received category
Property data information, the attribute data information is substituted into the regression curve, the value for predicting the default label may be implemented, from
And realize the dynamics simulation simulation of virtual train.
Support vector regression is the process of a supervised learning, exact according to history data set repetition training formation one
Training pattern directly can obtain the pre- bidding of demand using the training pattern in order to attribute data information certain known to the later period
Label, the entire model realization virtual analog of train, reduce the actual test amount of train.
With the continuous accumulation of sample, the training set in machine learning can be updated, to realize to the attribute number
It is believed that the continuous self study of breath and the default label, improves the accuracy of regression curve.
The embodiment of the present invention is by the support vector regression algorithm in machine learning, to the attribute data information and mark of train
Label are trained, and realize simulation and test to virtual train on emulation platform；Can reduce train field monitoring monitoring when
Between and manpower consumption, and improve efficiency and accuracy rate in entire test process.
Further, based on the above embodiment, before the basis presets tag extraction attribute data information, the side
Method further include: the setting default label, the default label specifically include trailer system link delay time, braking system chain
Road delay time and tractive force；Determine that the attribute data information, the attribute data information specifically include ATO speed, column
Distance away from nearest station of vehicle level, train, ATP speed, ATP speed limit, ATP alarm, level change the time, the ATO response time,
ATO draws acceleration and ATO braking acceleration.
Default label is set, the default label is to emulate the value that finally predicted.The default label can be with people
It is specified, the default label as described in specified according to emulation purpose.Label is preset described in the embodiment of the present invention specifically includes traction
System link delay time, braking system link delay time and tractive force, it is possible to understand that, the trailer system link prolongs
When time, braking system link delay time and tractive force be train operation in trailer system link delay time, braking system
System link delay time and tractive force.
In addition, can also be by obtaining important label using temperature figure tool, to realize the dynamics simulation to train.Benefit
It can be set with temperature figure for the label of simulation process and reference is provided.It is measured by test of many times, such as benefit in the embodiment of the present invention
It can know that the strong correlation attribute in the train dynamics parameter by selection converges to three key parameters substantially with temperature figure:
Trailer system link delay time, braking link delay time and tractive force；So selecting three keys in subsequent step
Label of the parameter as machine learning.Meanwhile if convergence result shows other protrusions by temperature map analysis repeatedly
Key parameter then should modify label through tradeoff at this time.
Determine that the attribute data information, the attribute data information are related to the default label according to default label
, influential on the result of the default label attribute data information.In three labels in the embodiment of the present invention, lead
Draw system link delay time and time when braking link delay time can be changed by level opens with corresponding ATO speed
The time difference that beginning changes thoroughly deserves, and tractive force can be obtained by tractive force model.To each in train travelling process
The research of institute's stress is divided into tractive force, running resistance, three kinds of brake force under operating status；Wherein running resistance and brake force influence
Tractive force size in entire train travelling process.
On the basis of the above embodiments, the embodiment of the present invention by setting label and attribute number relevant to label it is believed that
Breath, the realization for further machine learning algorithm provide premise, improve the reliability of emulation.
Further, based on the above embodiment, it is described according to the attribute data information demarcate the default label it
Before, the method also includes: the attribute data information is pre-processed and analyzed, is specifically included: removing abnormal attribute number
It is believed that breath, the attribute data information is normalized and is obtained the incidence relation of the attribute data information.
It removes abnormal attribute data information and refers to the abnormal data removed in the attribute data information.Remove abnormal attribute
Data information is to guarantee that the data reliability in subsequent step, abnormal data will affect subsequent regression algorithm to recurrence mould
The deduction of type；Since the data scale between the different attribute data information is different, in order to guarantee attribute data described in multiclass
Information does regression analysis in same data acquisition system, needs to be normalized, and otherwise can lose the lesser attribute of scale-value
Feature；The incidence relation of the attribute data information is obtained, it can be to provide reference by machine learning fit regression curve.
The embodiment of the present invention improves the effective of sample data by the way that attribute data information is pre-processed and analyzed
Property, improve the reliability and accuracy of curve matching.
Further, based on the above embodiment, described that pretreatment is carried out to the attribute data information and analyzes specific packet
It includes: the abnormal attribute data information being judged using box traction substation, and removes the abnormal attribute data information；Then it utilizes
The attribute data information is normalized in min-max standardized algorithm or Z-score standardized algorithm；Finally utilize
Temperature figure carries out strong correlation analysis to the attribute data information.
The abnormal attribute data information is judged using box traction substation, and removes the abnormal attribute data information；Due to
Data scale between the different abnormal attribute data informations is different, in order to guarantee that multi-class data is done in same data acquisition system
Regression analysis needs to be normalized, and otherwise can lose the lesser attributive character of scale-value, be standardized using min-max
Or normalized is completed in Z-score standardization, and the abnormal attribute data information is converted to nondimensional 0~1 value, with
Facilitate and compares；Strong correlation analysis finally is carried out to the attribute data information using temperature figure, finds strong correlation attribute to carry out
Further data analysis and processing.
Simulation software can be write completion by Python, be transported under 7 system of Windows by 2.7 interpreter of Python
Row, the simulation software can realize the function that the attribute data information is pre-processed and analyzed.
The embodiment of the present invention is by pre-processing attribute data information using box traction substation, normalization algorithm and temperature figure
With analysis, the validity of sample data is improved, improves the reliability and accuracy of curve matching.
Fig. 2 is the train dynamics simulation platform structure schematic diagram provided in an embodiment of the present invention based on machine learning.Such as
Shown in Fig. 2, the emulation platform includes attribute data data obtaining module 10, default label demarcating module 20, machine learning mould
Block 30 and simulation and prediction module 40, in which:
Attribute data data obtaining module 10 is specifically used for according to default tag extraction attribute data information, the attribute number
It is believed that breath is related to the default label；
The default label refers to by emulating the parameter to be predicted.Attribute data data obtaining module 10 is according to described pre-
Bidding label extract attribute data information relevant with the default label from the data board data of train, described to preset with described
Label correlation refers to that the value of the attribute data information has an impact to the value of the default label.Attribute data data obtaining module
After 10 extract the attribute data information, by the attribute data information preservation of extraction in local or database for emulation journey
Sequence is called.
Default label demarcating module 20 is specifically used for demarcating the default label according to the attribute data information；
The attribute data information is on the influential data of the value of default label tool, according to the kinetic simulation of train
Type or empirical equation obtain the relationship of the attribute data information Yu the default label, and then default label demarcating module 20
The default label is demarcated according to the attribute data information, namely the pre- bidding is calculated according to the attribute data information
Label.
Machine learning module 30 is specifically used for using the attribute data information and the default label as training set, base
It exercises supervision study in support vector regression algorithm, it is bent to obtain the recurrence that the default label is obtained by the attribute data information
Line；
When the attribute data information is various dimensions, machine learning module 30 is with the attribute data information and described pre-
Bidding label are input, are output with the default label, are trained using support vector regression algorithm；Wherein, as input
The attribute data information and the default label as output, obtain the regression curve for training；As input
The default label be used to supervision machine learning training process, during fitting obtains the regression curve, if root
The gap of the value of the predicted value and actual default label of the default label obtained according to the regression curve is beyond pre-
If threshold value then gives up respective sample.
When machine learning module 30 is analyzed using support vector regression, according to the attribute data information and the pre- bidding
The mass historical data of label is trained and is fitted, and it is bent to obtain the recurrence for obtaining the default label by the attribute data information
Line.
Simulation and prediction module 40 is specifically used for utilizing the regression curve, pre- according to the attribute data information prediction
It is marked with the value of label, realizes the dynamics simulation simulation of virtual train；
After the acquisition of machine learning module 30 obtains the regression curve of the default label by the attribute data information, emulation
Prediction module 40 substitutes into the regression curve for the new received attribute data information, by the attribute data information, can
To realize the value for predicting the default label, to realize the dynamics simulation simulation of virtual train.
The embodiment of the present invention is by the support vector regression algorithm in machine learning, to the attribute data information and mark of train
Label are trained, and realize simulation and test to virtual train on emulation platform；Can reduce train field monitoring monitoring when
Between and manpower consumption, and improve efficiency and accuracy rate in entire test process.
Further, based on the above embodiment, the emulation platform further include: default label setting module is specifically used for
The default label is set, when the default label specifically includes trailer system link delay time, braking system link delay
Between and tractive force；Attribute data information determination module is specifically used for determining the attribute data information, the attribute number it is believed that
Breath specifically includes ATO speed, train level, train distance, ATP speed, ATP speed limit, ATP alarm, level away from nearest station
Change time, ATO response time, ATO traction acceleration and ATO braking acceleration.
Default label setting module is used to set default label, and the default label is emulated and finally to be predicted
Value.The default label can be taking human as specified, the default label as described in specified according to emulation purpose.In addition, utilization can also be passed through
Temperature figure tool obtains important label, to realize the dynamics simulation to train.
Attribute data information determination module determines the attribute data information, the attribute data information according to default label
It is the attribute data information relevant to the default label, influential on the result of the default label.The present invention is real
It applies in three labels in example, when trailer system link delay time and braking link delay time can be changed by level
Time starts thoroughly deserving for the time difference changed with corresponding ATO speed, and tractive force can be obtained by tractive force model.
On the basis of the above embodiments, the embodiment of the present invention by setting label and attribute number relevant to label it is believed that
Breath, the realization for further machine learning algorithm provide premise, improve the reliability of emulation.
Further, based on the above embodiment, the emulation platform further include: pretreatment and analysis module are specifically used for
The attribute data information is pre-processed and is analyzed, comprising: remove abnormal attribute data information, to the attribute number it is believed that
Breath is normalized and obtains the incidence relation of the attribute data information.
It removes abnormal attribute data information and refers to the abnormal data removed in the attribute data information.Remove abnormal attribute
Data information is to guarantee that the data reliability in subsequent step, abnormal data will affect subsequent regression algorithm to recurrence mould
The deduction of type；Since the data scale between the different attribute data information is different, in order to guarantee attribute data described in multiclass
Information does regression analysis in same data acquisition system, needs to be normalized, and otherwise can lose the lesser attribute of scale-value
Feature；The incidence relation of the attribute data information is obtained, it can be to provide reference by machine learning fit regression curve.
The embodiment of the present invention improves the effective of sample data by the way that attribute data information is pre-processed and analyzed
Property, improve the reliability and accuracy of curve matching.
Further, based on the above embodiment, the pretreatment is specifically used for analysis module: judging institute using box traction substation
Abnormal attribute data information is stated, and removes the abnormal attribute data information；Then min-max standardized algorithm or Z- are utilized
The attribute data information is normalized in score standardized algorithm；Finally using temperature figure to the attribute data
Information carries out strong correlation analysis.
Pretreatment judges the abnormal attribute data information, and the abnormal attribute number using box traction substation with analysis module
It is believed that breath；Since the data scale between the different abnormal attribute data informations is different, in order to guarantee multi-class data same
Regression analysis is done in data acquisition system, needs to be normalized, otherwise can lose the lesser attributive character of scale-value, is utilized
Normalized is completed in min-max standardization or Z-score standardization, and the abnormal attribute data information is converted to dimensionless
0~1 value, compared with facilitating；Strong correlation analysis finally is carried out to the attribute data information using temperature figure, finds strong phase
Attribute is closed to carry out further data analysis and processing.
The embodiment of the present invention is by pre-processing attribute data information using box traction substation, normalization algorithm and temperature figure
With analysis, the validity of sample data is improved, improves the reliability and accuracy of curve matching.
Equipment provided in an embodiment of the present invention is for the above method, and concrete function can refer to above method process, this
Place repeats no more.
Fig. 3 is the structural schematic diagram of electronic equipment provided in an embodiment of the present invention.As shown in figure 3, electronic equipment 1 includes place
Manage device 301, memory 302 and bus 303.Wherein, the processor 301 and the memory 302 are complete by the bus 303
At mutual communication；The processor 301 is used to call the program instruction in the memory 302, to execute above-mentioned each side
Method provided by method embodiment, for example, according to default tag extraction attribute data information, the attribute data information with
The default label is related；The default label is demarcated according to the attribute data information；Using the attribute data information and
The default label is exercised supervision study based on support vector regression algorithm as training set, obtain by the attribute number it is believed that
Breath obtains the regression curve of the default label；It is pre- according to the attribute data information prediction using the regression curve
It is marked with the value of label, realizes the dynamics simulation simulation of virtual train.
The embodiment of the present invention discloses a kind of computer program product, and the computer program product is non-transient including being stored in
Computer program on computer readable storage medium, the computer program include program instruction, when described program instructs quilt
When computer executes, computer is able to carry out method provided by above-mentioned each method embodiment, for example, according to default label
Attribute data information is extracted, the attribute data information is related to the default label；It is demarcated according to the attribute data information
The default label；Using the attribute data information and the default label as training set, calculated based on support vector regression
Method exercises supervision study, obtains and obtains the regression curve of the default label by the attribute data information；Utilize the recurrence
Curve presets the value of label according to the attribute data information prediction, realizes the dynamics simulation simulation of virtual train.
The embodiment of the present invention provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage
Medium storing computer instruction, the computer instruction make the computer execute side provided by above-mentioned each method embodiment
Method, for example, according to default tag extraction attribute data information, the attribute data information is related to the default label；
The default label is demarcated according to the attribute data information；Using the attribute data information and the default label as instruction
Practice collection, exercised supervision study based on support vector regression algorithm, acquisition obtains the default label by the attribute data information
Regression curve；Using the regression curve, the value of label is preset according to the attribute data information prediction, is realized virtual
The dynamics simulation of train is simulated.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
The embodiments such as electronic equipment described above are only schematical, wherein it is described as illustrated by the separation member
Unit may or may not be physically separated, and component shown as a unit may or may not be object
Manage unit, it can it is in one place, or may be distributed over multiple network units.It can select according to the actual needs
Some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying wound
In the case where the labour for the property made, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers
It enables and using so that an electronic equipment (can be personal computer, server or the network equipment etc.) executes each embodiment
Or method described in certain parts of embodiment.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of train dynamics emulation mode based on machine learning characterized by comprising
According to default tag extraction attribute data information, the attribute data information is related to the default label；
The default label is demarcated according to the attribute data information；
Using the attribute data information and the default label as training set, exercised supervision based on support vector regression algorithm
Study obtains and obtains the regression curve of the default label by the attribute data information；
Using the regression curve, the value of label is preset according to the attribute data information prediction, realizes virtual train
Dynamics simulation simulation.
2. the method according to claim 1, wherein the basis preset tag extraction attribute data information it
Before, the method also includes:
The default label is set, the default label specifically includes the trailer system link delay time, braking system link prolongs
When time and tractive force；
Determine that the attribute data information, the attribute data information specifically include ATO speed, train level, train away from nearest
The distance at station, ATP speed, ATP speed limit, ATP alarm, level change the time, the ATO response time, ATO traction acceleration and
ATO braking acceleration.
3. the method according to claim 1, wherein described described pre- according to attribute data information calibration
Before bidding label, the method also includes:
The attribute data information is pre-processed and analyzed, is specifically included: removing abnormal attribute data information, to the category
Property data information is normalized and obtains the incidence relation of the attribute data information.
4. according to the method described in claim 3, it is characterized in that, described pre-processed and divided to the attribute data information
Analysis specifically includes:
The abnormal attribute data information is judged using box traction substation, and removes the abnormal attribute data information；Then it utilizes
The attribute data information is normalized in min-max standardized algorithm or Z-score standardized algorithm；Finally utilize
Temperature figure carries out strong correlation analysis to the attribute data information.
5. a kind of train dynamics emulation platform based on machine learning characterized by comprising
Attribute data data obtaining module, be specifically used for according to preset tag extraction attribute data information, the attribute number it is believed that
It ceases related to the default label；
Default label demarcating module is specifically used for demarcating the default label according to the attribute data information；
Machine learning module is specifically used for using the attribute data information and the default label as training set, based on branch
It holds vector regression algorithm to exercise supervision study, obtains and the regression curve of the default label is obtained by the attribute data information；
Simulation and prediction module is specifically used for utilizing the regression curve, be marked in advance according to the attribute data information prediction
The value of label realizes the dynamics simulation simulation of virtual train.
6. emulation platform according to claim 5, which is characterized in that the emulation platform further include:
Default label setting module is specifically used for setting the default label, and the default label specifically includes trailer system chain
Road delay time, braking system link delay time and tractive force；
Attribute data information determination module is specifically used for determining that the attribute data information, the attribute data information are specifically wrapped
Include distance away from nearest station of ATO speed, train level, train, ATP speed, ATP speed limit, ATP alarm, level change the time,
ATO response time, ATO traction acceleration and ATO braking acceleration.
7. emulation platform according to claim 5, which is characterized in that the emulation platform further include:
Pretreatment and analysis module, specifically for the attribute data information is pre-processed and analyzed, comprising: remove abnormal
Attribute data information, the association for the attribute data information being normalized and being obtained the attribute data information are closed
System.
8. emulation platform according to claim 7, which is characterized in that the pretreatment is specifically used for analysis module:
The abnormal attribute data information is judged using box traction substation, and removes the abnormal attribute data information；Then it utilizes
The attribute data information is normalized in min-max standardized algorithm or Z-score standardized algorithm；Finally utilize
Temperature figure carries out strong correlation analysis to the attribute data information.
9. a kind of electronic equipment, which is characterized in that including memory and processor, the processor and the memory pass through always
Line completes mutual communication；The memory is stored with the program instruction that can be executed by the processor, the processor tune
The method as described in Claims 1-4 is any is able to carry out with described program instruction.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt
The method as described in Claims 1-4 is any is realized when processor executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711296331.4A CN109918687B (en) | 2017-12-08 | 2017-12-08 | Train dynamics simulation method and simulation platform based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711296331.4A CN109918687B (en) | 2017-12-08 | 2017-12-08 | Train dynamics simulation method and simulation platform based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109918687A true CN109918687A (en) | 2019-06-21 |
| CN109918687B CN109918687B (en) | 2021-01-15 |
Family
ID=66956659
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711296331.4A Active CN109918687B (en) | 2017-12-08 | 2017-12-08 | Train dynamics simulation method and simulation platform based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109918687B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110513358A (en) * | 2019-08-28 | 2019-11-29 | 上海华兴数字科技有限公司 | Emulation mode, system, equipment and the storage medium of loading cylinders virtual prototype |
| CN111325462A (en) * | 2020-02-18 | 2020-06-23 | 中国铁道科学研究院集团有限公司 | Motor train unit auxiliary driving method and system |
| CN111860857A (en) * | 2020-04-15 | 2020-10-30 | 北京简单科技有限公司 | Student learning emotion distinguishing method and device based on intelligent learning environment |
| CN112287450A (en) * | 2019-07-09 | 2021-01-29 | 中车株洲电力机车研究所有限公司 | Train transmission control unit characteristic curve evaluation method, device, system and medium |
| CN112733448A (en) * | 2021-01-07 | 2021-04-30 | 北京理工大学 | Parameter self-learning double Q table combined agent establishing method for automatic train driving system |
| CN113722991A (en) * | 2021-08-24 | 2021-11-30 | 厦门大学 | Heat source parameter reverse identification method for laser direct deposition simulation |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104392071A (en) * | 2014-12-12 | 2015-03-04 | 北京交通大学 | High-speed train system security evaluation method based on complex network |
| CN104637334A (en) * | 2015-02-10 | 2015-05-20 | 中山大学 | Real-time predicting method for arrival time of bus |
| CN104765916A (en) * | 2015-03-31 | 2015-07-08 | 西南交通大学 | Dynamics performance parameter optimizing method of high-speed train |
| CN105404272A (en) * | 2015-11-24 | 2016-03-16 | 北京交控科技股份有限公司 | Automatic dynamic testing method for waking up of full-automatic driving train |
| CN105398438A (en) * | 2015-11-24 | 2016-03-16 | 株洲南车时代电气股份有限公司 | Track traffic train traction brake system and method |
| CN105404175A (en) * | 2015-11-24 | 2016-03-16 | 北京交控科技股份有限公司 | Stand-alone simulation system of vehicle-mounted equipment |
| CN105760658A (en) * | 2016-02-03 | 2016-07-13 | 华东交通大学 | High-speed train noise prediction method based on interval neural network |
| CN106202635A (en) * | 2016-06-28 | 2016-12-07 | 西安理工大学 | A kind of dynamic axle temperature Forecasting Methodology of bullet train based on multivariate regression models |
| CN106444421A (en) * | 2016-09-29 | 2017-02-22 | 南京理工大学 | Train traction-brake controller system of urban rail transit and working method of system |
| CN106777752A (en) * | 2016-12-30 | 2017-05-31 | 华东交通大学 | A kind of bullet train follows the trail of operation curve Optimal Setting method |
| CN106844949A (en) * | 2017-01-18 | 2017-06-13 | 清华大学 | A kind of training method for realizing the controllable two-way LSTM models of locomotive section |
| CN107423761A (en) * | 2017-07-24 | 2017-12-01 | 清华大学 | Feature based selects and the rail locomotive energy saving optimizing method of operating of machine learning |
-
2017
- 2017-12-08 CN CN201711296331.4A patent/CN109918687B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104392071A (en) * | 2014-12-12 | 2015-03-04 | 北京交通大学 | High-speed train system security evaluation method based on complex network |
| CN104637334A (en) * | 2015-02-10 | 2015-05-20 | 中山大学 | Real-time predicting method for arrival time of bus |
| CN104765916A (en) * | 2015-03-31 | 2015-07-08 | 西南交通大学 | Dynamics performance parameter optimizing method of high-speed train |
| CN105404272A (en) * | 2015-11-24 | 2016-03-16 | 北京交控科技股份有限公司 | Automatic dynamic testing method for waking up of full-automatic driving train |
| CN105398438A (en) * | 2015-11-24 | 2016-03-16 | 株洲南车时代电气股份有限公司 | Track traffic train traction brake system and method |
| CN105404175A (en) * | 2015-11-24 | 2016-03-16 | 北京交控科技股份有限公司 | Stand-alone simulation system of vehicle-mounted equipment |
| CN105760658A (en) * | 2016-02-03 | 2016-07-13 | 华东交通大学 | High-speed train noise prediction method based on interval neural network |
| CN106202635A (en) * | 2016-06-28 | 2016-12-07 | 西安理工大学 | A kind of dynamic axle temperature Forecasting Methodology of bullet train based on multivariate regression models |
| CN106444421A (en) * | 2016-09-29 | 2017-02-22 | 南京理工大学 | Train traction-brake controller system of urban rail transit and working method of system |
| CN106777752A (en) * | 2016-12-30 | 2017-05-31 | 华东交通大学 | A kind of bullet train follows the trail of operation curve Optimal Setting method |
| CN106844949A (en) * | 2017-01-18 | 2017-06-13 | 清华大学 | A kind of training method for realizing the controllable two-way LSTM models of locomotive section |
| CN107423761A (en) * | 2017-07-24 | 2017-12-01 | 清华大学 | Feature based selects and the rail locomotive energy saving optimizing method of operating of machine learning |
Non-Patent Citations (4)
| Title |
|---|
| NIKOLA MARKOVIC ET AL.: ""Analyzing passenger train arrival delays with support vector regression"", 《TRANSPORTATION RESEARCH PART C》 * |
| 王丽娟: ""高速列车位置计算模型与算法"", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 罗岩: ""预测控制在列车自动驾驶系统中的应用研究"", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 陈垚 等: ""基于支持向量回归的地铁牵引能耗预测"", 《系统工程理论与实践》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112287450A (en) * | 2019-07-09 | 2021-01-29 | 中车株洲电力机车研究所有限公司 | Train transmission control unit characteristic curve evaluation method, device, system and medium |
| CN112287450B (en) * | 2019-07-09 | 2022-11-08 | 中车株洲电力机车研究所有限公司 | Train transmission control unit characteristic curve evaluation method, device, system and medium |
| CN110513358A (en) * | 2019-08-28 | 2019-11-29 | 上海华兴数字科技有限公司 | Emulation mode, system, equipment and the storage medium of loading cylinders virtual prototype |
| CN111325462A (en) * | 2020-02-18 | 2020-06-23 | 中国铁道科学研究院集团有限公司 | Motor train unit auxiliary driving method and system |
| CN111860857A (en) * | 2020-04-15 | 2020-10-30 | 北京简单科技有限公司 | Student learning emotion distinguishing method and device based on intelligent learning environment |
| CN112733448A (en) * | 2021-01-07 | 2021-04-30 | 北京理工大学 | Parameter self-learning double Q table combined agent establishing method for automatic train driving system |
| CN113722991A (en) * | 2021-08-24 | 2021-11-30 | 厦门大学 | Heat source parameter reverse identification method for laser direct deposition simulation |
| CN113722991B (en) * | 2021-08-24 | 2023-10-03 | 厦门大学 | Heat source parameter reverse identification method for laser direct deposition simulation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109918687B (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109918687A (en) | A kind of train dynamics emulation mode and emulation platform based on machine learning | |
| Dong et al. | Characterizing driving styles with deep learning | |
| CN109783843B (en) | LSTM-based simulation system vehicle dynamics model speed prediction method | |
| CN104331442A (en) | Video classification method and device | |
| CN110362822A (en) | Text marking method, apparatus, computer equipment and storage medium for model training | |
| CN110458214B (en) | Driver replacement recognition method and device | |
| CN109545027A (en) | A kind of practical traning platform, crew's simulation training method and device | |
| CN112861706A (en) | Road state monitoring method, device, equipment and storage medium | |
| CN114328277A (en) | Software defect prediction and quality analysis method, device, equipment and medium | |
| CN110909907A (en) | Method and device for predicting fuel consumption of truck and storage medium | |
| CN110020190B (en) | Multi-instance learning-based suspicious threat index verification method and system | |
| CN118035411A (en) | Customer service voice quality inspection method, customer service voice quality inspection device, customer service voice quality inspection equipment and storage medium | |
| CN109388711A (en) | Method and device for clustering log streams | |
| CN117390407A (en) | Fault identification method, system, medium and equipment of substation equipment | |
| CN112308434A (en) | Traffic safety risk assessment method and system | |
| Wang et al. | A driver abnormal behavior warning method based on isolated forest algorithm. | |
| CN114818484A (en) | Training method of driving environment model and prediction method of driving environment information | |
| CN113609777A (en) | Vehicle behavior analysis method and device based on vehicle-mounted sensing equipment | |
| WO2022179759A1 (en) | Generating unknown-unsafe scenarios, improving automated vehicles, computer system | |
| CN113592341A (en) | Measurement loss function, sector complexity evaluation method and system | |
| Dabbous et al. | Bringing Intelligence to the Edge for Structural Health Monitoring. The Case Study of the Z24 Bridge | |
| Ding et al. | Finite mixture of the hidden Markov model for driving style analysis | |
| EP4293334A1 (en) | Crash test device, apparatus, method and computer program for controlling a crash test device and for training a neural network | |
| CN117975178B (en) | Taxi track data analysis method based on big data analysis | |
| CN118035939B (en) | Confidence coefficient acquisition method for perception target and automatic driving planning control method |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100070 12 / F, block a, yard 1, South Automobile Museum Road, Fengtai District, Beijing Applicant after: Tonghao Urban Rail Transit Technology Co., Ltd. Address before: 100073 11 Floor, Block D, No. 1 South Road, Fengtai Auto Museum, Beijing Applicant before: Beijing Tonghao National Railway Urban Rail Technology Co., Ltd. |
|
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Ma Chong Inventor after: Wang Luyi Inventor after: Peng Chaoyang Inventor after: Zhao Anan Inventor after: Zhang Chen Inventor after: Tian Yuan Inventor before: Ma Chong Inventor before: Wang Luyi Inventor before: Han Huijie Inventor before: Peng Chaoyang Inventor before: Zhang Chen Inventor before: Zhao Anan |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |