CN117302293B - Wheel diameter value prediction method and device, electronic equipment and storage medium - Google Patents
Wheel diameter value prediction method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN117302293B CN117302293B CN202311022245.XA CN202311022245A CN117302293B CN 117302293 B CN117302293 B CN 117302293B CN 202311022245 A CN202311022245 A CN 202311022245A CN 117302293 B CN117302293 B CN 117302293B
- Authority
- CN
- China
- Prior art keywords
- wheel diameter
- diameter value
- value
- prediction
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 100
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 41
- 238000010801 machine learning Methods 0.000 claims abstract description 13
- 238000012937 correction Methods 0.000 claims description 35
- 230000006870 function Effects 0.000 claims description 35
- 238000004590 computer program Methods 0.000 claims description 14
- 230000002159 abnormal effect Effects 0.000 claims description 13
- 230000001419 dependent effect Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 description 22
- 230000008569 process Effects 0.000 description 20
- 239000013598 vector Substances 0.000 description 19
- 230000008859 change Effects 0.000 description 10
- 238000005259 measurement Methods 0.000 description 10
- 238000002922 simulated annealing Methods 0.000 description 8
- 238000005299 abrasion Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000003062 neural network model Methods 0.000 description 6
- 230000001133 acceleration Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 4
- 230000004927 fusion Effects 0.000 description 4
- 238000009825 accumulation Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 239000005433 ionosphere Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 239000005436 troposphere Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L15/00—Indicators provided on the vehicle or train for signalling purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Probability & Statistics with Applications (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention relates to the technical field of rail transit, and provides a wheel diameter value prediction method, a device, electronic equipment and a storage medium, wherein the method comprises the steps of firstly obtaining a wheel diameter value sequence of a target train on a historical road section; and then inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model. According to the method, the wheel diameter value of each road section can be subjected to iterative prediction by using the prediction model, so that dynamic prediction of the wheel diameter value is realized, the wheel diameter value of a target train in the current section can be predicted in advance, the wheel diameter value prediction efficiency is improved, and the timeliness and accuracy of using the wheel diameter value of the train in the next section are improved. According to the method, the initial model is constructed by using a trend prediction algorithm based on machine learning, so that the prediction model can be guaranteed to have the capability of predicting the wheel diameter value of the target train in the current section. Meanwhile, the method provides a determination step of the value of the super parameter of the initial model, so that the prediction performance of the prediction model can be improved.
Description
Technical Field
The present invention relates to the field of rail traffic technologies, and in particular, to a wheel diameter value prediction method, a device, an electronic apparatus, and a storage medium.
Background
At present, in urban subways, a speed sensor and transponder combined positioning mode is generally adopted to position a train, and the accumulated running distance of the train is calculated through an odometer, but due to the influence of wheel diameter abrasion of the train, the train has positioning errors, and the errors are accumulated continuously along with the running of the train. Although the positioning accuracy of the train can be effectively improved by densely arranging the transponders, the construction and maintenance costs are increased, and the updating of the line configuration is not facilitated.
The wheel diameter value is an important parameter for train positioning, calculation of motor rotation speed and linear speed and conversion of force and moment. When the error of the wheel diameter value setting is overlarge or the error occurs, certain influences are caused on the traction, the braking system and the constant speed control. In the running process of the train, the deviation of wheel diameter values can lead to the running speed and running distance calculated by the automatic train protection subsystem (Automatic Train Protection, ATP) through the speed sensor and software logic to be different from the actual running speed and running distance, so that the train cannot be positioned accurately.
The method is a rail transit train wheel diameter correction method based on multi-sensor information fusion, and when all sensors work normally, the method obtains the optimal estimation of the train wheel diameter value through multi-sensor information filtering fusion and completes wheel diameter correction.
However, when the sensor has problems, the method causes interruption of calculation of wheel diameter values, unsuccessful wheel diameter correction, adopts default data, and can not eliminate the influence of various external random interferences and measurement noise on a measurement result only by the sensor, so that inaccurate calculation precision of the path value can be caused, inaccurate calculation position and speed of a train can be caused, the train can be continuously braked in a traction manner in the running process, and the same acceleration is not adopted for stable running, thereby influencing the experience of passengers and the traction abrasion of the train.
Disclosure of Invention
The invention provides a wheel diameter value prediction method, a wheel diameter value prediction device, electronic equipment and a storage medium, which are used for solving the defects in the prior art.
The invention provides a wheel diameter value prediction method, which comprises the following steps:
acquiring a wheel diameter value sequence of a target train on a historical road section;
inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
According to the wheel diameter value prediction method provided by the invention, the initial wheel diameter value in the wheel diameter value sequence is determined based on the wheel diameter correction transponder when the target train passes through the wheel diameter correction transponder at the initial road section.
According to the wheel diameter value prediction method provided by the invention, the sample wheel diameter value is determined based on the following steps:
Determining an initial sample wheel diameter value of each road section based on wheel diameter correction transponders arranged on each road section on the sample travel path;
filling abnormal values and missing values in the initial sample wheel diameter values, and determining the sample wheel diameter values.
According to the wheel diameter value prediction method provided by the invention, the abnormal value is obtained based on the judgment of the box diagram.
According to the wheel diameter value prediction method provided by the invention, the prediction model is a self-attention model.
According to the wheel diameter value prediction method provided by the invention, the wheel diameter value sequence is input into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model, and the method comprises the following steps:
Inputting the wheel diameter value sequence to a coding module of the self-attention model to obtain a characteristic representation sequence output by the coding module;
and inputting the characteristic representation sequence to a decoding module of the self-attention model to obtain the wheel diameter value of the target train at the current road section, which is output by the decoding module.
The invention also provides a wheel diameter value prediction device, which comprises:
The acquisition module is used for acquiring a wheel diameter value sequence of the target train in the historical road section;
The prediction module is used for inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
According to the wheel diameter value predicting device provided by the invention, the initial wheel diameter value in the wheel diameter value sequence is determined based on the wheel diameter correction transponder when the target train passes through the wheel diameter correction transponder at the initial road section.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the wheel diameter value prediction method according to any one of the above when executing the computer program.
The invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a wheel diameter value prediction method as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a wheel diameter value prediction method as described in any one of the above.
The invention provides a wheel diameter value prediction method, a device, electronic equipment and a storage medium, wherein the method firstly obtains a wheel diameter value sequence of a target train on a historical road section; and then inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model. According to the method, the wheel diameter value of each road section can be subjected to iterative prediction by using the prediction model, so that dynamic prediction of the wheel diameter value is realized, the wheel diameter value prediction is not required to be carried out after a train passes through the wheel diameter correction transponder, the wheel diameter value of a target train in the current section can be predicted in advance, the wheel diameter value prediction efficiency is improved, and the timeliness and the accuracy of using the wheel diameter value in the next section of the train are improved. According to the method, the initial model is constructed by using a trend prediction algorithm based on machine learning, so that the prediction model can be guaranteed to have the capability of predicting the wheel diameter value of the target train in the current section. Meanwhile, the method provides a determination step of the value of the super parameter of the initial model, so that the determined value of the super parameter is the optimal value, and the prediction performance of the prediction model can be improved. The method can not cause interruption of the calculation process of the wheel diameter value even when a sensor on the running path of the train is in a problem, and can eliminate the influence of various external random interferences and measurement noise on a measurement result, so that the obtained wheel diameter value is more accurate and has higher precision, thereby providing guarantee for accurate calculation of the position and the speed of the train, reducing the frequency of traction braking in the running process of the train, adopting the same acceleration to stably run, improving the experience of passengers and reducing the traction abrasion of the train. In addition, due to accurate calculation of wheel diameter values, errors generated in train positioning can be reduced, and continuous accumulation of the errors along with train running is reduced.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious to those skilled in the art that other drawings can be obtained according to these drawings without inventive effort.
FIG. 1 is a schematic flow chart of a wheel diameter value prediction method provided by the invention;
FIG. 2 is a schematic diagram of the overall architecture of a prediction model in the wheel diameter value prediction method provided by the invention;
FIG. 3 is a schematic software architecture diagram of the wheel diameter value prediction method provided by the invention;
FIG. 4 is a complete flow of the wheel diameter value prediction method implemented by the self-attention model based on the meta-heuristic algorithm provided by the invention;
FIG. 5 is a schematic diagram of the wheel diameter value predicting device according to the present invention;
Fig. 6 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The features of the invention "first", "second" and the like in the description and in the claims may be used for the explicit or implicit inclusion of one or more such features. In the description of the invention, unless otherwise indicated, the meaning of "a plurality" is two or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/", generally means that the associated object is an "or" relationship.
In the prior art, when the wheel diameter value is predicted, a rail transit train wheel diameter correction method based on multi-sensor information fusion is adopted, the method aims at the problem that the wheel diameter is reduced due to gradual abrasion of train wheels in the running process, so that the speed measurement positioning precision of a wheel shaft speed sensor is affected, a train combined positioning system is formed by analyzing the error characteristics of a positioning sensor and adopting the wheel shaft speed sensor, an accelerometer and a Doppler speed measuring radar, and the advantage complementation is carried out by utilizing different positioning principles of different sensors, so that the train positioning is realized; and combining with a Kalman filtering theory, a wheel diameter value calculation method based on Kalman filtering is provided. According to the method, when each sensor works normally, optimal estimation of the train wheel diameter value is obtained through multi-sensor information filtering fusion, and wheel diameter correction is completed. However, when the sensor has problems, the method causes interruption of calculation of wheel diameter values, unsuccessful wheel diameter correction, adopts default data, and can not eliminate the influence of various external random interferences and measurement noise on a measurement result only by the sensor, so that inaccurate calculation precision of the path value can be caused, inaccurate calculation position and speed of a train can be caused, the train can be continuously braked in a traction manner in the running process, and the same acceleration is not adopted for stable running, thereby influencing the experience of passengers and the traction abrasion of the train.
The other method is a train automatic wheel diameter correction method based on the speed of a global positioning system (Global Positioning System, GPS), the method is a method for automatically calculating the wheel diameter value of each shaft according to the speed of the GPS and the rotating speed of a motor of each shaft, an inertial measurement device is not needed to be additionally arranged, line data are acquired in advance, only the GPS device is needed to be arranged, the line speed of train operation is acquired through the GPS device, a central control unit acquires the rotating speed of the motor and the GPS speed through a multifunctional train bus (Multifunction Vehicle Bus, MVB) and an Ethernet, and the wheel diameter value of each shaft can be calculated according to the principle that the displacement is equal in a certain time. The method saves cost and is convenient to verify the wheel diameter value result in a simulation mode. However, since the signal source of the GPS is that the satellite is sent to the GPS signal receiving device of the train after a certain policy calculation. The communication period of the control system for acquiring the motor rotation speed is very short, and when the motor rotation speed changes greatly, the GPS speed updating period is slower, so that the calculation result of the wheel diameter value can deviate greatly. Meanwhile, the calculation of the wheel diameter value has very high precision requirement on the GPS linear speed, and a plurality of factors influencing the GPS signal quality are as follows: such as atmospheric layers (ionosphere and troposphere delays for GPS signals), multipath effects (GPS signals reflected by other surfaces into the receiver antenna), etc., which cause in most cases deviations in GPS speed from the true speed of the train, and increase in correction time and difficulty, resulting in reduced accuracy in wheel diameter value calculation.
Based on the above, the embodiment of the invention provides a wheel diameter value prediction method.
From statistical analysis of wheel diameter values, the wheel diameter values are characterized by:
Periodicity: since the process of running trains from the start station to the end station is periodic, each train runs on a fixed line periodically, and the line condition is unchanged, the abrasion of the wheel diameter values is periodic.
Stability: the wheel diameter correction transponder is delivered after being strictly tested, the wheel diameter value reversely deduced by the wheel diameter correction transponder is almost stable in the process of stable running of the train under the conditions of daily environment and stable passenger number, and fluctuation can not occur in a certain time, but if the variation of the daily environment and the passenger number is too large or unknown conditions occur, certain influence can be generated on the stability of the wheel diameter value data.
Noisy, non-linear: the wheel diameter value reversely deduced by the wheel diameter correction transponder when the train runs on track also depends on environmental change, if the long-term wheel diameter correction transponder has problems in deduced value because of unpredictable factors, the data such as the running speed, the running distance and the like calculated by ATP are noisy.
Fig. 1 is a schematic flow chart of a wheel diameter value prediction method provided in an embodiment of the present invention, as shown in fig. 1, the method includes:
s1, acquiring a wheel diameter value sequence of a target train in a historical road section;
s2, inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
Specifically, in the wheel diameter value prediction method provided in the embodiment of the present invention, the execution body is a wheel diameter value prediction device, and the device may be configured in a computer, where the computer may be a local computer or a cloud computer, and the local computer may be a computer, a tablet, or the like, and is not limited herein specifically.
Step S1 is executed first, and a wheel diameter value sequence of a target train in a historical road section is obtained. The target train may be a train requiring wheel diameter value prediction during its traveling, and the traveling path of the target train is a path between a start station and an end station, and may include a plurality of road segments, each road segment corresponding to a wheel diameter value.
The wheel diameter value of the current road section can be calculated through iterative calculation of the wheel diameter value of the previous road section. That is, for a current road section other than a starting road section of the target train on the running path, in order to predict the wheel diameter value thereof, it is necessary to obtain a sequence of wheel diameter values of a history road section through which the target train has run before the current road section. The historical road sections can comprise one or more, and the wheel diameter values of all the historical road sections form a wheel diameter value sequence according to the sequence of the historical road sections. For example, if the history road includes n total road segments, the road diameter value sequence includes n total road diameter values, which may be represented as (X 1,…,Xn).
It is understood that the wheel diameter value sequence belongs to a periodic and nonlinear time sequence, and the wheel diameter value in the wheel diameter value sequence has a certain trend.
Then, step S2 is executed, in which the wheel diameter value of the current road segment, i.e. the n+1th road segment, is predicted by means of the prediction model, and at this time, the sequence of wheel diameter values of the historical road segments may be input into the prediction model, and the wheel diameter value X n+1 of the target train on the current road segment may be obtained and output from the prediction model. The prediction model works in an iterative mode, namely after the wheel diameter value of the current road section is determined, the current road section is changed into a historical road section along with continuous running of the train, the wheel diameter value sequence is updated to (X 1,…,Xn,Xn+1), and then the next road section, namely the (n+2) th road section, can be used as the current road section, and the wheel diameter value X n+2 of the next road section is predicted. The prediction model can iterate the wheel diameter value prediction process until the train reaches the terminal station.
The prediction model can be obtained by training an initial model by using sample wheel diameter values of each road section of a sample train on a sample running path. For example, when training the initial model, for any road section except the initial road section on the sample driving path, the sample wheel diameter value of any road section can be input into the initial model to obtain the wheel diameter value prediction result of the next road section of any road section output by the initial model, then the value of the loss function is calculated through the sample wheel diameter value of the next road section and the wheel diameter value prediction result, and when the value of the loss function converges, the trained initial model, namely the prediction model, is obtained. Here, the loss function may be selected as needed, for example, a minimum mean square error may be selected as the loss function.
It should be noted that, the initial model may be constructed based on a trend prediction algorithm of machine learning, so that the prediction model may be guaranteed to have the capability of predicting the wheel diameter value of the target train in the current section. The trend prediction algorithm may include a BP algorithm, a statistical regression algorithm, a graph neural network algorithm, etc., and accordingly the initial model may be a BP neural network model, a statistical regression model, a graph neural network model, etc.
In addition, the value of the hyper-parameter of the initial model can be determined based on a meta-heuristic algorithm, and the determined value of the hyper-parameter can be made to be the optimal value, so that the prediction performance of the prediction model can be improved. The meta-heuristic algorithm can search globally to a certain extent to find an approximate solution of the optimal solution. The core of the meta-heuristic algorithm is exploration and utilization. Wherein, since the optimal solution may exist at any position of the whole search space, the exploration process needs to explore the whole search space as much as possible. The utilization process needs to utilize effective information as much as possible, and the correlation among the optimal solutions is utilized to gradually adjust, so that the optimal solutions are slowly searched from the initial solutions.
The meta heuristic algorithm comprises a tabu search algorithm, a simulated annealing algorithm, a genetic algorithm, an ant colony optimization algorithm, a particle swarm optimization algorithm, an artificial fish swarm algorithm, an artificial bee colony algorithm, an artificial neural network algorithm and the like.
The wheel diameter value prediction method provided by the embodiment of the invention comprises the steps of firstly, acquiring a wheel diameter value sequence of a target train in a historical road section; and then inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model. According to the method, the wheel diameter value of each road section can be subjected to iterative prediction by using the prediction model, so that dynamic prediction of the wheel diameter value is realized, the wheel diameter value prediction is not required to be carried out after a train passes through the wheel diameter correction transponder, the wheel diameter value of a target train in the current section can be predicted in advance, the wheel diameter value prediction efficiency is improved, and the timeliness and the accuracy of using the wheel diameter value in the next section of the train are improved. According to the method, the initial model is constructed by using a trend prediction algorithm based on machine learning, so that the prediction model can be guaranteed to have the capability of predicting the wheel diameter value of the target train in the current section. Meanwhile, the method provides a determination step of the value of the super parameter of the initial model, so that the determined value of the super parameter is the optimal value, and the prediction performance of the prediction model can be improved. The method can not cause interruption of the calculation process of the wheel diameter value even when a sensor on the running path of the train is in a problem, and can eliminate the influence of various external random interferences and measurement noise on a measurement result, so that the obtained wheel diameter value is more accurate and has higher precision, thereby providing guarantee for accurate calculation of the position and the speed of the train, reducing the frequency of traction braking in the running process of the train, adopting the same acceleration to stably run, improving the experience of passengers and reducing the traction abrasion of the train. In addition, due to accurate calculation of wheel diameter values, errors generated in train positioning can be reduced, and continuous accumulation of the errors along with train running is reduced.
On the basis of the above embodiment, according to the wheel diameter value prediction method provided by the embodiment of the present invention, the initial wheel diameter value in the wheel diameter value sequence is determined based on the wheel diameter correction transponder when the target train passes through the wheel diameter correction transponder at the initial road section.
Specifically, a wheel diameter correction transponder is arranged at a starting road section of a running path of a target train, the wheel diameter correction transponder can be used for calculating a wheel diameter value of the starting road section, and the wheel diameter value of the starting road section is an initial wheel diameter value in a wheel diameter value sequence, so that an accurate actual initial wheel diameter value can be provided for iterative prediction of a prediction model.
On the basis of the above embodiment, the wheel diameter value prediction method provided in the embodiment of the present invention, where the value of the super parameter is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
Specifically, the value of the super parameter is determined by adopting a meta heuristic algorithm based on the following steps:
Firstly, a preset number of target sample wheel diameter values in the sample wheel diameter values can be randomly selected, the preset number can be set according to the needs, and the method is not particularly limited herein, so long as the accurate determination of the values can be ensured. And constructing and obtaining a target model by utilizing the super parameters.
Then, an objective function f (x) is constructed by taking the super parameter as an independent variable and the information amount obtained from the target sample wheel diameter value by the target model as a dependent variable. Where x is a hyper-parameter and y=f (x) is a dependent variable. The amount of information may be represented by entropy.
Finally, solving the independent variable value which enables the dependent variable value of the objective function f (x) to be the maximum as the value of the super-parameter.
In the solving process, if an old solution x i exists in the objective function currently, a new solution x j can be found near x i to update the old solution until the termination condition is met, and the solution x * is the value of the super parameter.
For the old solution x i, the objective function value is f (x i), and for the new solution x j, the objective function value is f (x j). If f (x j)>f(xi), then prove that the new solution x j is better than x i, then replace the old solution x i with the new solution x j; otherwise, an acceptance probability p e [0,1] is set, and the smaller the distance d (x i,xj) between f (x i) and f (x j), the larger the acceptance probability p.
Different meta-heuristics employ different strategies, such as hill climbing, if the new solution x j is not better than x i, the old solution is not replaced with the new solution, i.e. the acceptance probability p=0. For the simulated annealing algorithm, the distance d (x i,xj) is shown in equation (1).
Since the acceptance probability p is inversely proportional to d (x i,xj), it can be obtained
Here, the simulated annealing algorithm is taken as an example, and further analysis is performed.
In the early operation stage of the simulated annealing algorithm, a larger search range is needed, so that the simulated annealing algorithm can better acquire a new solution, and a larger acceptance probability p is needed at the moment;
in the later stage of the simulated annealing algorithm, a relatively accurate solution needs to be searched within a certain range in a narrowing range, and a smaller acceptance probability p needs to be adopted at the moment.
For this case, a time-dependent control variable c t is introduced, ensuring that the acceptance probability p is inversely proportional to the control variable c t, and that the control variable c t is incremented over time step t.
The simulated annealing algorithm assumes that the initial temperature T 0 = 100, and the temperature T t+1 for time step t+1 drops as shown in equation (2) compared to the temperature T t for time step T. Wherein, a is the parameter of (0, 1), which is set to 0.95 in the embodiment of the invention.
Tt+1＝aTt； (2)
The calculation of the control variable c t, which is inversely proportional to T t, is shown in equation (3).
Thus, the final result of the acceptance probability p is shown in formula (4).
Finally, the flow of the simulated annealing algorithm to find the maximum of the objective function f (x) is as follows:
first, a solution x k is randomly generated, and the corresponding objective function value f (x k) is calculated.
Second, a solution x k+1 is randomly generated around x k, and the corresponding objective function value f (x k+1) is calculated. If f (x k+1)>f(xk), then x k is replaced with solution x k+1. Otherwise, the probability of accepting the solution x k+1 is calculated as shown in equation (5). Then generating a random number r (b) using the random number seed b, and accepting the solution x k+1 to replace x k if r (b) < p; otherwise, the original solution x k is maintained.
Third, the second step is repeated continuously at the temperature T t of the time step r until the number of iterations E t is set at the current temperature. The time and temperature are then updated according to equation (2). The second step is repeated continuously according to the new time and temperature until the termination condition is met.
The termination condition may be one of three types: the algorithm reaches the final iteration number E; (2) The temperature has fallen to a specified threshold T *; (3) The optimal solution x * is found and then is iterated for several times without change.
On the basis of the above embodiment, the wheel diameter value prediction method provided in the embodiment of the present invention, where the sample wheel diameter value is determined based on the following steps:
Determining an initial sample wheel diameter value of each road section based on wheel diameter correction transponders arranged on each road section on the sample travel path;
filling abnormal values and missing values in the initial sample wheel diameter values, and determining the sample wheel diameter values.
Specifically, the wheel diameter value data generated in the running process of the train may be abnormal due to factors such as environmental influence, component faults and the like, so that the performance of the obtained prediction model can be trained to generate adverse effects. Therefore, when determining the sample wheel diameter value for training the initial model, the wheel diameter correction transponder of each road section on the sample driving path can be utilized to determine the initial sample wheel diameter value of each road section, and then the initial sample wheel diameter value is preprocessed, namely, abnormal values and missing values in the initial sample wheel diameter value are filled.
The abnormal value and the missing value can be filled by adopting an average value, so that the normal sample wheel diameter value can be conveniently and quickly determined.
On the basis of the above embodiment, the wheel diameter value prediction method provided in the embodiment of the present invention, the abnormal value is determined based on the box chart.
Specifically, since the wheel diameter value data does not always strictly conform to the normal distribution, the abnormal value in the initial sample wheel diameter value can be determined by using the box chart, and the abnormal value in the initial sample wheel diameter value can be determined quickly and accurately.
The BP neural network model prediction has the biggest defects of low calculation speed and high requirement on the original data volume; the statistical regression model needs to find a linear or nonlinear function to fit the change process of the original sequence, so that the implementation process is complex; the expansibility of the graph neural network model is limited, and the deep network is easy to generate characteristic smoothness, so that the discriminant of the obtained characteristic representation is reduced.
Based on the above embodiments, the wheel diameter value prediction method provided in the embodiment of the present invention is based on the above embodiments, and the prediction model is a self-attention model.
Specifically, the self-attention model essentially models the correlation between different parts in the whole input, thereby avoiding the influence of the error accumulation error of the previous t-1 time slices in the process of predicting the t time slice of the existing model based on the deep learning, avoiding the limitation of the model to a square or rectangular receptive field (RECEPTIVE FIELD), determining the shape and type of the receptive field of the model, and improving the accuracy and efficiency of the model. Specifically, the self-attention model multiplies all input vectors by a learnable parameter matrix, respectively, generates a query vector q i, a key vector k i, a value vector v i for each vector a i, and calculates the weight w i of the vector a i in the entire self-attention model output by calculating the similarity of the query vector k i and the key vector v i. Then given n input vectors, the output value is
The basic idea of the self-attention model is to treat the data sequence formed by the predicted object over time as a random sequence, which is approximately described by the self-attention model. After the model is constructed, future information can be predicted according to historical information and current information.
If the prediction model is a self-attention model, the model can be considered as a self-attention model realized based on a meta-heuristic algorithm, and the relation between the sample wheel diameter value of a certain road section and the sample wheel diameter values of other road sections of the train on the sample driving path can be calculated through the self-attention model, so that the relation between the sample wheel diameter values of all the road sections is modeled.
The self-attention model can better overcome the defect of poor prediction precision of a statistical-based prediction model in nonlinear prediction, solve the problem of low efficiency of a BP neural network model and the problem of low expansibility of a graph neural network model, has good adaptability, and can show good prediction performance when in linear and nonlinear fields.
On the basis of the above embodiment, the wheel diameter value prediction method provided in the embodiment of the present invention inputs the wheel diameter value sequence to a prediction model, and obtains the wheel diameter value of the target train at the current road section output by the prediction model, including:
Inputting the wheel diameter value sequence to a coding module of the self-attention model to obtain a characteristic representation sequence output by the coding module;
and inputting the characteristic representation sequence to a decoding module of the self-attention model to obtain the wheel diameter value of the target train at the current road section, which is output by the decoding module.
Specifically, the overall architecture of the prediction model is shown in fig. 2, including an encoding module and a decoding module. The coding module is particularly based on a self-Attention mechanism, and adds a Multi-Head Attention (Multi-Head Attention) structure and a Feed Forward neural network (Feed Forward). The multi-head attention structure can pay attention to different positions in the input wheel diameter value sequence, and the expression capability of self-attention on each road section in the wheel diameter value sequence is enhanced.
The feedforward neural network can introduce nonlinear transformation through a function activating mode, so that the expression capacity of a prediction model is further enhanced, and the fitting effect is enhanced. The coding module maps the input wheel diameter value sequence (X 1,…,Xn) into a continuous characteristic representation sequence (Z 1,…,Zn), and finally the decoding module predicts the wheel diameter value of the current road section according to the characteristic representation sequence and acquires the content of important attention so as to generate the wheel diameter value of the current road section.
The specific self-attention calculation is shown in formula (6):
Wherein Q, K, V are three proofs, namely a Query vector (Query), a Key vector (Key), and a Value vector (Value), respectively, the attention weight is obtained by calculating the similarity between the Query vector Q and the Key vector K, and then the final attention Value is obtained on the Value vector by the attention weight.
The overall prediction process of the prediction model is as follows:
First, Z is a characteristic representation sequence obtained after embedding the wheel diameter value sequence X, and Z i and Z j are characteristic representations of wheel diameter values of two road sections. Multiplying Z by weight matrices W Q,WK and W V yields the corresponding query vector Q, key vector K, and value vector V.
Second, a self-attention score value is calculated, which indicates the degree of attention to the wheel diameter values of other road segments when the wheel diameter value of a certain road segment is encoded, for example, a score value of the wheel diameter value Z j of other road segments to the wheel diameter value of the road segment is calculated for the wheel diameter value X i of the i-th road segment. The wheel diameter value X i for the i-th road segment is Q i·Ki, and the wheel diameter value X j for the other road segments is Q i·Kj.
Third, dividing the score value byAnd then, calculating a softmax to obtain the correlation of the wheel diameter value X j of the other road sections to the wheel diameter value X i of the i-th road section.
Fourth, the value vector V and the value obtained after softmax are multiplied and added, so that the wheel diameter value of the current section of the target train output by the prediction model is obtained.
On the basis of the above embodiment, the software structure of the wheel diameter value prediction method provided in the embodiment of the present invention is shown in fig. 3, and includes: the device comprises an initialization module, a data acquisition module, a data recording module, a synchronization module, a prediction model and a wheel diameter value output module.
When the target train passes through the wheel diameter correction transponder of the initial road section, the data acquisition module acquires the wheel diameter value of the initial road section determined by the wheel diameter correction transponder and records the value through the data recording module, the synchronous module inputs the value acquired by the data acquisition module into the prediction model, the wheel diameter value of the target train on the current road section is obtained through the prediction model, and the wheel diameter value of the target train on the current road section is stored into the data recording module through the wheel diameter value output module.
And the numerical value acquired by the data acquisition module is synchronously input into the prediction model by the synchronization module, the wheel diameter value of the target train at the next road section is obtained by the prediction model, and the wheel diameter value of the target train at the next road section is stored into the data recording module by the wheel diameter value output module, so that the cycle is carried out until the target train runs to the terminal.
According to the wheel diameter value prediction method provided by the embodiment of the invention, the accuracy of wheel diameter value prediction is improved in a dynamic prediction mode.
The wheel diameter value of the target train is continuously worn along with the running of the target train, the whole change trend of the wheel diameter value is monotonously reduced trend change, the reduced trend change of the wheel diameter value is changed along with the change of the running time and the running mileage, the method is based on time sequence change, an analysis method based on a prediction model is adopted, and the change trend of the wheel diameter value in a future period of time is predicted according to the change trend of the wheel diameter value in a previous period of time.
The following is the complete flow steps of the wheel diameter value prediction method implemented by the self-attention model based on the meta-heuristic algorithm, and the processing flow is shown in fig. 4:
1) The method comprises the steps of collecting initial sample wheel diameter values of all road sections of a sample train on a sample running path, preprocessing the sample wheel diameter values, and selecting a preset number of target sample wheel diameter values in the preprocessed sample wheel diameter values.
2) And randomly acquiring an initial solution of the super parameter by using a meta heuristic algorithm according to the preprocessed target sample wheel diameter value, and setting a termination condition as that the optimal solution is unchanged for 30 continuous iterations.
3) A new solution is randomly generated near the initial solution, and whether the initial solution is to be updated with the new solution is judged by calculating the distance between the new solution and the initial solution, as shown in formula (5).
4) Returning to the value of the super parameter of the initial self-attention model when the termination condition is met, and updating the temperature and the time to continuously solve when the condition is not met.
5) The initial self-attention model is trained to obtain a self-attention model.
6) And inputting a wheel diameter value sequence of the target train on the historical road section, and outputting a wheel diameter value X n+1 of the target train on the current road section through the trained self-attention model.
7) X n+1 is used as an input value of the next wheel diameter value prediction, and the step 6) is circularly carried out, so that the wheel diameter value is dynamically corrected in the running process of the train.
As shown in fig. 5, on the basis of the above embodiment, an embodiment of the present invention provides a wheel diameter value prediction apparatus, including:
The acquiring module 51 is configured to acquire a wheel diameter value sequence of a target train on a historical road section;
The prediction module 52 is configured to input the wheel diameter value sequence to a prediction model, and obtain a wheel diameter value of the target train at a current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
On the basis of the above embodiment, the wheel diameter value predicting device provided in the embodiment of the present invention, where the initial wheel diameter value in the wheel diameter value sequence is determined based on the wheel diameter correction transponder when the target train passes through the wheel diameter correction transponder at the initial road section.
On the basis of the above embodiment, the wheel diameter value predicting device provided in the embodiment of the present invention further includes a sample wheel diameter value determining module, configured to:
Determining an initial sample wheel diameter value of each road section based on wheel diameter correction transponders arranged on each road section on the sample travel path;
filling abnormal values and missing values in the initial sample wheel diameter values, and determining the sample wheel diameter values.
On the basis of the above embodiment, the wheel diameter value prediction device provided in the embodiment of the present invention, the abnormal value is determined based on the box chart.
On the basis of the above embodiment, the wheel diameter value prediction device provided in the embodiment of the present invention is characterized in that the prediction model is a self-attention model.
On the basis of the above embodiment, the wheel diameter value prediction device provided in the embodiment of the present invention is the prediction module, which is configured to:
Inputting the wheel diameter value sequence to a coding module of the self-attention model to obtain a characteristic representation sequence output by the coding module;
and inputting the characteristic representation sequence to a decoding module of the self-attention model to obtain the wheel diameter value of the target train at the current road section, which is output by the decoding module.
Specifically, the functions of each module in the wheel diameter value prediction device provided in the embodiment of the present invention are in one-to-one correspondence with the operation flows of each step in the above method embodiment, and the achieved effects are consistent.
Fig. 6 illustrates a physical schematic diagram of an electronic device, as shown in fig. 6, which may include: processor (Processor) 610, communication interface (Communications Interface) 620, memory (Memory) 630, and communication bus 640, wherein Processor 610, communication interface 620, memory 630 complete communication with each other through communication bus 640. The processor 610 may call logic instructions in the memory 630 to perform the wheel diameter value prediction method provided in the above embodiments, the method comprising: acquiring a wheel diameter value sequence of a target train on a historical road section; inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model; the prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps: selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters; taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function; solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
Further, the logic instructions in the memory 630 may be implemented in the form of software functional units and stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of executing the wheel diameter value prediction method provided in the above embodiments, the method comprising: acquiring a wheel diameter value sequence of a target train on a historical road section; inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model; the prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined as follows: selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters; taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function; solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the wheel diameter value prediction method provided in the above embodiments, the method comprising: acquiring a wheel diameter value sequence of a target train on a historical road section; inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model; the prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps: selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters; taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function; solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A wheel diameter value prediction method, comprising:
acquiring a wheel diameter value sequence of a target train on a historical road section;
inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
2. The wheel diameter value prediction method according to claim 1, wherein an initial wheel diameter value in the sequence of wheel diameter values is determined based on the wheel diameter correction transponder of the target train when the starting road section passes the wheel diameter correction transponder.
3. The wheel diameter value prediction method according to claim 1, wherein the sample wheel diameter value is determined based on the steps of:
Determining an initial sample wheel diameter value of each road section based on wheel diameter correction transponders arranged on each road section on the sample travel path;
filling abnormal values and missing values in the initial sample wheel diameter values, and determining the sample wheel diameter values.
4. The wheel diameter value prediction method according to claim 3, wherein the abnormal value is determined based on a box map.
5. The wheel diameter value prediction method according to any one of claims 1 to 4, characterized in that the prediction model is a self-attention model.
6. The wheel diameter value prediction method according to claim 5, wherein inputting the sequence of wheel diameter values into a prediction model to obtain the wheel diameter value of the target train at the current road section output by the prediction model comprises:
Inputting the wheel diameter value sequence to a coding module of the self-attention model to obtain a characteristic representation sequence output by the coding module;
and inputting the characteristic representation sequence to a decoding module of the self-attention model to obtain the wheel diameter value of the target train at the current road section, which is output by the decoding module.
7. A wheel diameter value predicting apparatus, comprising:
The acquisition module is used for acquiring a wheel diameter value sequence of the target train in the historical road section;
The prediction module is used for inputting the wheel diameter value sequence into a prediction model to obtain the wheel diameter value of the target train at the current road section, which is output by the prediction model;
The prediction model is obtained by training an initial model based on sample wheel diameter values of all road sections of a sample train on a sample running path, the initial model is constructed based on a trend prediction algorithm of machine learning, and the value of the super parameter of the initial model is determined based on the following steps:
Selecting a preset number of target sample wheel diameter values in the sample wheel diameter values, and constructing a target model based on the super parameters;
Taking the super parameter as an independent variable, and taking the information quantity obtained by the target model from the target sample wheel diameter value as an independent variable to construct a target function;
Solving the independent variable value with the maximum dependent variable value of the objective function as the value of the super-parameter.
8. The wheel diameter value prediction device according to claim 7, wherein an initial wheel diameter value in the sequence of wheel diameter values is determined based on the wheel diameter correction transponder of the target train when the starting road section passes the wheel diameter correction transponder.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the wheel diameter value prediction method of any of claims 1-6 when the computer program is executed.
10. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the wheel diameter value prediction method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311022245.XA CN117302293B (en) | 2023-08-14 | 2023-08-14 | Wheel diameter value prediction method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311022245.XA CN117302293B (en) | 2023-08-14 | 2023-08-14 | Wheel diameter value prediction method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117302293A CN117302293A (en) | 2023-12-29 |
| CN117302293B true CN117302293B (en) | 2024-06-07 |
Family
ID=89296112
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311022245.XA Active CN117302293B (en) | 2023-08-14 | 2023-08-14 | Wheel diameter value prediction method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117302293B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110861678A (en) * | 2019-11-27 | 2020-03-06 | 通号城市轨道交通技术有限公司 | Train wheel diameter checking method and device |
| CN112590874A (en) * | 2020-12-21 | 2021-04-02 | 交控科技股份有限公司 | Train wheel diameter correction method and device |
| CN113734237A (en) * | 2021-11-03 | 2021-12-03 | 比亚迪股份有限公司 | Method for determining wheel diameter of vehicle, storage medium and electronic device |
| CN113799849A (en) * | 2021-11-16 | 2021-12-17 | 比亚迪股份有限公司 | Train wheel diameter calibration method, vehicle-mounted controller, train and storage medium |
| CN114117687A (en) * | 2021-12-08 | 2022-03-01 | 中车青岛四方机车车辆股份有限公司 | Method and system for building and predicting life prediction model of key parts of wheel set |
| WO2022141213A1 (en) * | 2020-12-30 | 2022-07-07 | 中南大学 | Gene prediction method and system for fault of autonomous rail rapid transit vehicle in smart city |
| CN114971034A (en) * | 2022-05-31 | 2022-08-30 | 中国铁路广州局集团有限公司广州动车段 | Analysis algorithm for predicting remaining life of wheel |
| CN115130728A (en) * | 2022-05-31 | 2022-09-30 | 通号城市轨道交通技术有限公司 | Method and device for predicting train wheel diameter calibration time |
| CN115593463A (en) * | 2022-10-25 | 2023-01-13 | 国能朔黄铁路发展有限责任公司（Cn） | Vehicle wheel diameter correction method and device, storage medium and electronic equipment |
-
2023
- 2023-08-14 CN CN202311022245.XA patent/CN117302293B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110861678A (en) * | 2019-11-27 | 2020-03-06 | 通号城市轨道交通技术有限公司 | Train wheel diameter checking method and device |
| CN112590874A (en) * | 2020-12-21 | 2021-04-02 | 交控科技股份有限公司 | Train wheel diameter correction method and device |
| WO2022141213A1 (en) * | 2020-12-30 | 2022-07-07 | 中南大学 | Gene prediction method and system for fault of autonomous rail rapid transit vehicle in smart city |
| CN113734237A (en) * | 2021-11-03 | 2021-12-03 | 比亚迪股份有限公司 | Method for determining wheel diameter of vehicle, storage medium and electronic device |
| CN113799849A (en) * | 2021-11-16 | 2021-12-17 | 比亚迪股份有限公司 | Train wheel diameter calibration method, vehicle-mounted controller, train and storage medium |
| CN114117687A (en) * | 2021-12-08 | 2022-03-01 | 中车青岛四方机车车辆股份有限公司 | Method and system for building and predicting life prediction model of key parts of wheel set |
| CN114971034A (en) * | 2022-05-31 | 2022-08-30 | 中国铁路广州局集团有限公司广州动车段 | Analysis algorithm for predicting remaining life of wheel |
| CN115130728A (en) * | 2022-05-31 | 2022-09-30 | 通号城市轨道交通技术有限公司 | Method and device for predicting train wheel diameter calibration time |
| CN115593463A (en) * | 2022-10-25 | 2023-01-13 | 国能朔黄铁路发展有限责任公司（Cn） | Vehicle wheel diameter correction method and device, storage medium and electronic equipment |
Non-Patent Citations (2)
| Title |
|---|
| 基于灰色理论的列车组合定位轮径校准方法研究;刘江;蔡伯根;王剑;唐涛;;铁道学报;20110515(05);第57-59页 * |
| 改进的灰度预测在列车轮径校正中的研究;崔超;张亚东;郭进;;铁道标准设计;20160131(01);第141-143页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117302293A (en) | 2023-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108986470B (en) | Travel time prediction method for optimizing LSTM neural network by particle swarm optimization | |
| US10936940B2 (en) | Arrangement and method for predicting road friction within a road network | |
| CN110738855B (en) | Road traffic flow condition prediction method in data sparse time period | |
| CN105091889A (en) | Hotspot path determination method and hotspot path determination equipment | |
| CN113291321A (en) | Vehicle track prediction method, device, equipment and storage medium | |
| CN111121791A (en) | Optimization method of hidden Markov model in map matching and GPS positioning method | |
| CN117141520B (en) | Real-time track planning method, device and equipment | |
| CN110909953A (en) | Parking position prediction method based on ANN-LSTM | |
| CN114510778A (en) | Track irregularity prediction method based on hybrid intelligent optimization LSTM | |
| CN112818036A (en) | Electric vehicle remaining mileage estimation method and device based on cloud big data | |
| CN115496201A (en) | Train accurate parking control method based on deep reinforcement learning | |
| Bajic et al. | Road roughness estimation using machine learning | |
| CN115457764A (en) | Road section traffic density estimation method, device and medium based on vehicle track data | |
| CN113342474B (en) | Method, equipment and storage medium for predicting customer flow and training model | |
| CN117302293B (en) | Wheel diameter value prediction method and device, electronic equipment and storage medium | |
| CN112534446A (en) | Electronic control device and neural network updating system | |
| CN111160594B (en) | Method and device for estimating arrival time and storage medium | |
| CN115691140B (en) | Analysis and prediction method for space-time distribution of automobile charging demand | |
| CN117104310A (en) | Virtual marshalling control method and system based on data-driven predictive control | |
| CN117116064A (en) | Passenger delay minimization signal control method based on deep reinforcement learning | |
| CN115130728A (en) | Method and device for predicting train wheel diameter calibration time | |
| CN113902427A (en) | Method, device and equipment for determining estimated arrival time and storage medium | |
| CN117523821B (en) | System and method for predicting vehicle multi-mode driving behavior track based on GAT-CS-LSTM | |
| CN116542528B (en) | Traffic demand satisfaction analysis method and device | |
| CN117979229A (en) | Advanced aiming precision optimization method, device, equipment and storage medium |
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 |