CN116611864A - Urban rail transit train speed and passenger flow prediction method based on RNN - Google Patents
Urban rail transit train speed and passenger flow prediction method based on RNN Download PDFInfo
- Publication number
- CN116611864A CN116611864A CN202310587350.1A CN202310587350A CN116611864A CN 116611864 A CN116611864 A CN 116611864A CN 202310587350 A CN202310587350 A CN 202310587350A CN 116611864 A CN116611864 A CN 116611864A
- Authority
- CN
- China
- Prior art keywords
- train
- speed
- data
- passenger flow
- current
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000003062 neural network model Methods 0.000 claims abstract description 6
- 238000013528 artificial neural network Methods 0.000 claims description 22
- 125000004122 cyclic group Chemical group 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 15
- 238000013145 classification model Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 13
- 238000007621 cluster analysis Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000009499 grossing Methods 0.000 claims description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 2
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000012937 correction Methods 0.000 claims description 2
- 230000005674 electromagnetic induction Effects 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 239000013598 vector Substances 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 abstract description 5
- 238000003064 k means clustering Methods 0.000 abstract 1
- 238000011160 research Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 5
- 230000007613 environmental effect Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L27/00—Central railway traffic control systems; Trackside control; Communication systems specially adapted therefor
- B61L27/60—Testing or simulation
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61C—LOCOMOTIVES; MOTOR RAILCARS
- B61C17/00—Arrangement or disposition of parts; Details or accessories not otherwise provided for; Use of control gear and control systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Mechanical Engineering (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Game Theory and Decision Science (AREA)
- Transportation (AREA)
- Automation & Control Theory (AREA)
- Probability & Statistics with Applications (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The invention discloses a prediction method of urban rail transit train speed and passenger flow based on RNN, firstly, collecting voltage and current of each part of the train and known ramp and curve of subway lines and subway station data, and preprocessing the collected data; secondly, calculating power data of different intervals according to train set operation information and collected information; carrying out interval identification by using a K-means clustering method power curve, and storing information of the same interval as a sample data set and numbering; according to sample power data sets of different intervals, a train passenger flow prediction model is established and prediction is carried out; and finally, based on a sample data set and the train passenger flow volume in the same interval, establishing a circulating neural network model to realize real-time tracking and prediction of the train speed. The method has the advantages of being scientific, accurate and reliable, and can avoid the running faults of the train in advance, thereby providing a reference for ensuring the safe and stable running of the train.
Description
Technical Field
The invention relates to the technical field of urban rail transit, in particular to an RNN-based urban rail transit train speed and passenger flow prediction method.
Background
With the rapid development of Chinese economy and the continuous acceleration of urban process, urban population is rapidly increased, and the demand of people for urban traffic is continuously increased. Urban rail transit is used as a novel traffic mode, and has the advantages of large traffic volume, high punctual rate, safety, reliability, energy conservation, environmental protection and the like, so that the urban rail transit becomes the primary choice for people to travel. In order to improve punctuality and safety in the running process of urban rail transit trains and realize accurate stopping while urban rail transit is rapidly developed, a speed curve in the running process of the trains needs to be tracked and predicted, and the method has become one of hot spots in the research of the field of rail transit.
The existing research on the speed tracking prediction of the subway train still has certain problems, and has the following defects:
(1) Most of existing subway train simulations are to perform dynamic modeling on a train, solve the positions and accelerations of key points of the train by using a heuristic algorithm, and obtain a speed curve by using a train kinematics model, but the generation process of the method for acquiring the key point data of the train is complex, and accurate modeling is difficult, so that the existing model may not be capable of performing accurate tracking prediction work on the train;
(2) The existing subway train tracking prediction method is mostly based on the sensor information fusion technology for sampling, and the problems that the universality is not strong, equipment needs to be added and the like possibly exist, but the mathematical methods such as calculating the vehicle-mounted speed curve by adopting a second-order Bezier curve are not implemented on an operation line, and the universality is yet to be verified;
(3) The existing train power online rolling prediction method based on the time sequence long-short-period memory network does not consider the influence of passenger flow on a research result in the research, so that the operation management requirements of subway companies and the traveling experience of passengers are less considered, and the obtained optimization scheme lacks practical value.
Disclosure of Invention
In order to solve the defects in the background art, the invention provides the urban rail transit train speed and passenger flow prediction method based on the RNN, which adopts a method of a cyclic neural network RNN (RecurrentNeural Network), utilizes the thought of combining wireless sampling, real-time calculation and prediction, has the potential of being directly applied to the actual site of subway train operation for speed tracking prediction, and can accurately, reliably and scientifically predict the train speed.
The invention adopts the following technical scheme for solving the technical problems:
an urban rail transit train speed and passenger flow prediction method based on RNN comprises the following steps:
step 1: based on a data acquisition system arranged at the bottom of a train, the current sensor is utilized to acquire real-time data of traction current, auxiliary current and braking resistance current of the subway train and send the data to a server for storage; carrying out on-line testing on empty load, passenger carrying, front and rear double-train, multiple-train and the like on the train respectively;
step 2: calculating the instantaneous power in each section of the train according to the running information of the train set through the traction converter current, the auxiliary converter current and the contact net voltage data of the pantograph of the train;
step 3: classifying the sections by using a K-means cluster analysis model, taking the calculated train instantaneous power in each section as an input quantity, taking a train running section number as a classification label, carrying out network training based on K-means cluster analysis by adjusting network parameter values to obtain a section classification model, storing power information, current information, line information and passenger flow information of the same section as a sample data set, numbering, and correcting an abnormal section by using a feedback correction method;
step 4: based on a sample data set stored in the same interval, instantaneous power, current information and line information are taken as input, train passenger flow and speed are taken as output, a circulating neural network model is built, and real-time tracking and prediction of a train speed curve based on the circulating neural network are realized.
Furthermore, the data acquisition system in step 1 is installed at the bottom of a train, the system adopts an electromagnetic induction principle to sample, and adopts a wireless mode to transmit, so as to reduce the environmental influence of network fluctuation and train running conditions on data, and the operation of preprocessing comprises the steps of removing invalid data such as train leaving, warehousing, stopping, turning back and the like, and adopting average smoothing and normalization operation on power data.
Further, the calculating of the train instantaneous power in the step 2 specifically includes:
determining an instantaneous power calculation formula of the train according to the running information of the studied train; if the measured data object is n dr Motor-saving vehicle with n aux The vehicles of the section trailer consist trains, the instantaneous power p (t) is:
p(t)＝u(t)×(n dr i dr (t)-n aux i aux (t))
wherein n is dr I is the number of traction transformers on the train dr To draw the current of the converter, n aux I is the number of auxiliary transformers on the train aux U is the pantograph voltage, which is the auxiliary current transformer current.
Further, the K-means cluster analysis interval classification model in the step 3 specifically includes:
firstly, extracting features of a power curve, creating a certain initial centroid, calculating the distance between each centroid and a data point, distributing the distances to the nearest cluster, and training a classification model by using the serial numbers of all intervals as classification labels to obtain an interval classification model.
Further, after the current running section of the train is identified by using the K-means cluster analysis classification model, the cyclic neural network is established by taking the power curve, ramp information and curve information as inputs of the network and taking the train speed and the passenger flow as outputs for each section on the line, and error back propagation is carried out, the model is trained, and then tracking prediction is carried out on the train speed; the specific steps of the step 4 are as follows:
step 41: determining adjustable parameters such as the number of input layer nodes, the number of hidden layer nodes, the number of layers of the cyclic neural network, an activation function, a learning rate, an optimizer, a loss function, iteration times, batch sizes and the like of the cyclic neural network structure, and automatically updating the weight and bias parameters after selecting the network structure and other super parameters;
step 42: when a sample data set is generated, setting the current moment as a center, and selecting data points as input in a sliding window mode; adding ramp and curve data as input vectors, and outputting a current moment speed value and a passenger flow value; adopting a circulating neural network based on an Adam optimizer, wherein the input layer node is a sliding window dimension plus 2, two hidden layers are arranged, the hidden layer nodes are 128, the layer number of the circulating neural network is set to be 2, and the activation function is a leakage-ReLu function;
step 43: and inputting the preprocessed power curve, braking current information, ramp, curve data and passenger flow information into the established prediction model, training the cyclic neural network model, gradually reducing a loss function in the training process, and finally converging the model parameters to stable values.
Further, when tracking and predicting the train speed of the whole subway line, comparison and analysis are needed to be carried out on tracking and predicting results of different algorithms, and if the tracking and predicting errors meet the requirements, a tracking and predicting result curve of the final speed is output.
Compared with the prior art, the technical scheme has the following beneficial effects:
(1) The invention uses the neural network to predict the train operation energy consumption under different running time intervals based on the train operation data acquired and measured by the data acquisition board system, utilizes the thought of combining wireless sampling, real-time calculation and prediction, has the potential of being directly applied to the actual site of subway train operation for speed tracking prediction, and has smaller error of tracking prediction;
(2) According to the invention, a section classification algorithm based on K-means cluster analysis is introduced into the step of tracking and predicting the train speed, the prediction of the train full line running speed is decomposed into sections for research, the fault tolerance is higher, the problems of dimension disasters and the like are avoided, and the prediction precision is improved;
(3) In the process of optimizing the interval running time, the invention comprehensively considers the line data of train running, the subway passenger flow data and the operation management requirements, and has more practical application value.
Drawings
FIG. 1 is a flow chart of a method for predicting urban rail transit train speed and passenger flow based on RNN according to the present invention;
FIG. 2 is a graph comparing the predicted result and the actual result of the passenger flow of a certain subway line in China;
FIG. 3 is a diagram of a recurrent neural network for each interval of the present invention;
FIG. 4 is a graph comparing measured data of a certain section of a certain subway line in China with a speed tracking prediction result;
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
Fig. 1 is a flowchart of a method for predicting speed and passenger flow of an RNN-based urban rail transit train of the present invention. Referring to fig. 1, a method for predicting speed and passenger flow of an RNN-based urban rail transit train specifically includes the following steps:
step one: the method is characterized in that a certain domestic subway line is taken as a study object, a data acquisition system is independently installed outside a train to acquire data such as pantograph contact net voltage, current transformer current and the like in the running process of the train, train speed data for comparison are obtained from vehicle-mounted records of subway companies, and passenger flow data, line ramps and curve data are also provided by the subway companies. The train data of 40 days are collected in the experiment, wherein the data of one month from 7 months of 2021 to 12 days of 8 months of 22 days is used as a training set, and the data of all days of 10 months of 2022 is used as a test set.
Step two: the research object adopts a standard B-type vehicle, the train is formed by grouping 6 vehicles of 4 motor vehicles and 2 trailers, each motor vehicle is provided with a traction converter, direct current from a contact net is converted into three-phase alternating current, and the starting, braking and speed regulation of a traction motor are realized to drive the train to run. Each trailer is provided with an auxiliary converter, and the load is train auxiliary equipment, so that the instantaneous total power of the train is as follows:
p(t)＝u(t)×(4i dr (t)-2i aux (t))
wherein i is dr To draw the current of the converter, i aux U is the pantograph voltage, which is the auxiliary current transformer current.
Step three: the preprocessing operation is performed on the acquired data, because the actual measurement data is greatly influenced by external factors and has obvious fluctuation, and therefore, the processing of the data is necessary.
Further, the preprocessing in the third step includes:
step 3.1: firstly, eliminating data with obvious sampling errors, eliminating data during the period of ex-warehouse, stop-warehouse and turn-back of a train in the morning and the evening, and only retaining train data in the running process;
step 3.2: secondly, for the situation that the current data is influenced by environmental factors such as train motion state and the like, the calculated power curve jitter frequency is high due to high data jitter frequency, the actual running situation of the vehicle cannot be accurately reflected, and the curve after invalid data is removed is subjected to mean value smoothing;
step 3.3: when the actual measurement data are considered, the total power obtained by superposing the traction power and the braking power in the train braking stage is found when the traction power and the braking resistance power are compared, and the change trend of the total power can more reflect the speed curve of the train braking stage, so that the total power is selected as the input more reasonably.
Fig. 2 is a smoothed result of the train instantaneous power curve.
Step four: and establishing a K-means cluster analysis interval classification model, and carrying out interval identification according to the calculated train power curve. Because the power curves of all the sections have larger shape difference, the section where the train is currently located can be positioned through the power curve. When the classification model is built, in order to solve the problem of low classification precision caused by dimension disasters, the power curve is subjected to feature extraction, and the interval number is used as a classification label. Finally, a classification model is built according to the steps.
Step five: after any section of current train power curve is given, firstly, a section classification model is used for identifying a current section, and a neural network model is called from a subsequent section according to an identification result to track and predict subsequent train speed and passenger flow. Further, step five, a cyclic neural network prediction model is constructed to input power into a network through a sliding window function, ramp parameters of a current position of a train at a current moment and curve parameters of the current position of the train are added as input quantity of the network, expected output quantity is speed of the train at the moment and passenger flow quantity of the current section, and parameters of the network are updated by using an error back propagation algorithm, and the method specifically comprises the following steps:
step 5.1: determining each parameter of the cyclic neural network to be established, including the number of network layers, the number of neurons at each layer, the number of layers of the cyclic neural network, the type of activation function and the like, and determining training parameters, including a parameter updating method, an updating step length and a termination condition;
step 5.2: and analyzing the trains according to intervals according to the training set data. Calculating the instantaneous power of the train according to the preprocessed voltage and current data in each section, and looking up a table according to the sampling interval to obtain ramp parameters and curve parameters when the train is positioned at the time point, wherein the speed curve and section passenger flow of the train are used as the output of a training set, and the circulating neural network structure is shown in fig. 3;
step 5.3: after the input and output of the cyclic neural network are determined, the root mean square error is used as a loss function, the error is counter-propagated, and the training is continuously carried out until the termination condition is achieved;
step 5.4: firstly, carrying out data preprocessing on train operation data in a test set, then carrying out interval numbering on a power curve by using an interval classification model, then calling a circulating neural network model in a training set according to the number, substituting the input of the test set including the power curve, current data, line information and passenger flow information in the interval of the train into a network established in the interval, so as to carry out tracking prediction on the speed of the train and the passenger flow, and drawing a prediction result and measured data for qualitative analysis.
Finally, to quantitatively evaluate the tracked predicted train speed and the actual speed, relevant evaluation indexes such as RMSE, MAE and R are defined 2 The performance of the proposed predictive model is evaluated.
Wherein y is i In order to predict the speed of the vehicle,for the actual speed, n is the total number of predicted points, and the coefficient R is determined 2 The value range of (2) is [0,1 ]]The closer it is to 1, the more accurate it represents the predicted result.
As shown in FIG. 4, in one section of a subway line, the speed tracking prediction is performed by using the method provided by the invention, and the prediction curve is observed qualitatively, so that the train speed can be effectively tracked and predicted.
In addition, as shown in tables 1 and 2, experimental data show that the prediction method (Kmeans-RNN) of urban rail transit train speed and passenger flow based on RNN according to the embodiment of the invention can effectively control related error measurement indexes compared with the traditional RNN prediction model. Although the measured data cannot completely fit the full line speed of the train due to the length of the subway train line, the change of the train speed and the maximum speed of the train can be effectively reflected. By combining the results, the method provided by the embodiment of the invention can effectively realize speed tracking prediction of a certain subway train in running, has good stability, high prediction precision and high calculation speed, and can meet the actual requirements.
TABLE 1 passenger flow tracking prediction results
TABLE 2 speed tracking prediction results
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions may be made without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (6)
1. The method for predicting the speed and the passenger flow of the urban rail transit train based on the RNN is characterized by comprising the following steps of:
step 1: based on a data acquisition system arranged at the bottom of a train, the current sensor is utilized to acquire real-time data of traction current, auxiliary current and braking resistance current of the subway train and send the data to a server for storage; carrying out on-line testing on empty load, passenger carrying, front and rear double-train, multiple-train and the like on the train respectively;
step 2: calculating the instantaneous power in each section of the train according to the running information of the train set through the traction converter current, the auxiliary converter current and the contact net voltage data of the pantograph of the train;
step 3: classifying the sections by using a K-means cluster analysis model, taking the calculated train instantaneous power in each section as an input quantity, taking a train running section number as a classification label, carrying out network training based on K-means cluster analysis by adjusting network parameter values to obtain a section classification model, storing power information, current information, line information and passenger flow information of the same section as a sample data set, numbering, and correcting an abnormal section by using a feedback correction method;
step 4: based on a sample data set stored in the same interval, instantaneous power, current information and line information are taken as input, train passenger flow and speed are taken as output, a cyclic neural network RNN model is built, and real-time tracking and prediction of a train speed curve based on the cyclic neural network RNN are realized.
2. The method for predicting the speed and the passenger flow of the urban rail transit train based on the RNN according to claim 1, wherein the data acquisition system in the step 1 is arranged at the bottom of the train, adopts an electromagnetic induction principle to sample, adopts a wireless mode to transmit, and adopts average smoothing and normalization operations on power data to eliminate invalid data such as train leaving, warehousing, stopping, turning back and the like in order to reduce the influence of network fluctuation and the environment of train running conditions on the data.
3. The method for predicting the speed and the passenger flow of the urban rail transit train based on the RNN according to claim 1, wherein the calculating the instantaneous power in each section of the train in the step 2 specifically comprises:
determining an instantaneous power calculation formula of the train according to the running information of the studied train; if the measured data object is n dr Motor-saving vehicle with n aux The vehicles of the section trailer consist trains, the instantaneous power p (t) is:
p(t)＝u(t)×(n dr i dr (t)-n aux i aux (t))
wherein n is dr I is the number of traction transformers on the train dr To draw the current of the converter, n aux I is the number of auxiliary transformers on the train aux U is the pantograph voltage, which is the auxiliary current transformer current.
4. The method for predicting the speed and the passenger flow of the urban rail transit train based on the RNN according to claim 1, wherein the K-means cluster analysis in the step 3 specifically comprises the following steps:
and extracting the characteristics of the power curve, creating a certain initial centroid, calculating the distance between each centroid and a data point, distributing the closest cluster, and training a classification model by using the serial numbers of all intervals as classification labels to obtain an interval classification model.
5. The method for predicting the speed and the passenger flow of the urban rail transit train based on the RNN according to claim 1, wherein the real-time tracking prediction of the train speed curve based on the RNN of the cyclic neural network in step 4 comprises the following steps: after the K-means cluster analysis classification model is used for identifying the current running interval of the train, taking power curves, ramp information and curve information as network inputs and taking train speed and passenger flow as outputs aiming at each interval on the line, establishing a circulating neural network model, carrying out error back propagation, training the model and further carrying out tracking prediction on the train speed; the method comprises the following specific steps:
step 41: determining the number of input layer nodes, the number of hidden layer nodes, the number of layers of the cyclic neural network, an activation function, a learning rate, an optimizer, a loss function, iteration times, batch sizes and other adjustable parameters of the cyclic neural network RNN structure, and automatically updating the weight and bias parameters after selecting the network structure and other super parameters;
step 42: when a sample data set is generated, setting the current moment as a center, and selecting data points as input in a sliding window mode; adding ramp and curve data as input vectors, and outputting a current moment speed value and a passenger flow value; adopting a circulating neural network RNN based on an Adam optimizer, wherein the input layer node is a sliding window dimension plus 2, two hidden layers are arranged, the hidden layer nodes are 128, the layer number of the circulating neural network is set to be 2, and the activation function is a leakage-ReLu function;
step 43: and inputting the preprocessed power curve, braking current information, ramp, curve data and passenger flow information into the established model, training the RNN model of the cyclic neural network, gradually reducing a loss function in the training process, and finally converging the model parameters to stable values.
6. The RNN-based urban rail transit train speed and passenger flow volume prediction method according to claim 1, wherein when train speed tracking prediction is performed on the whole subway line, comparison analysis is required to be performed on tracking prediction results of different algorithms, and if tracking prediction errors meet requirements, a tracking prediction result curve of a final speed is output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310587350.1A CN116611864A (en) | 2023-05-24 | 2023-05-24 | Urban rail transit train speed and passenger flow prediction method based on RNN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310587350.1A CN116611864A (en) | 2023-05-24 | 2023-05-24 | Urban rail transit train speed and passenger flow prediction method based on RNN |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116611864A true CN116611864A (en) | 2023-08-18 |
Family
ID=87677721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310587350.1A Pending CN116611864A (en) | 2023-05-24 | 2023-05-24 | Urban rail transit train speed and passenger flow prediction method based on RNN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116611864A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117829378A (en) * | 2024-03-04 | 2024-04-05 | 华东交通大学 | Track traffic energy consumption prediction method based on space-time data |
-
2023
- 2023-05-24 CN CN202310587350.1A patent/CN116611864A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117829378A (en) * | 2024-03-04 | 2024-04-05 | 华东交通大学 | Track traffic energy consumption prediction method based on space-time data |
| CN117829378B (en) * | 2024-03-04 | 2024-05-14 | 华东交通大学 | Track traffic energy consumption prediction method based on space-time data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111832225B (en) | Method for constructing driving condition of automobile | |
| CN109086946B (en) | Method for predicting emission of polluted gas of conventional energy and new energy public transport vehicle | |
| Chen et al. | Soft computing methods applied to train station parking in urban rail transit | |
| CN110555476B (en) | Intelligent vehicle lane change track prediction method suitable for man-machine hybrid driving environment | |
| CN116611864A (en) | Urban rail transit train speed and passenger flow prediction method based on RNN | |
| CN112198799B (en) | High-speed train parking control method and system based on deep learning | |
| CN114510778A (en) | Track irregularity prediction method based on hybrid intelligent optimization LSTM | |
| CN116541790A (en) | New energy vehicle health assessment method and device based on multi-feature fusion | |
| Liu et al. | Braking process identification of high-speed trains for automatic train stop control | |
| CN112381027A (en) | Wheel polygon wave depth estimation method based on train axle box vertical acceleration signal | |
| CN110210326B (en) | Online train identification and speed estimation method based on optical fiber vibration signals | |
| CN114580262B (en) | Lithium ion battery health state estimation method | |
| Cunillera et al. | Real-time train motion parameter estimation using an unscented Kalman filter | |
| CN111898822B (en) | Charging load interval prediction method based on multi-correlation-day scene generation | |
| CN115222160B (en) | Rail transit traction load prediction method based on measured big data | |
| CN110245461A (en) | A kind of urban track traffic traction load modeling method, system and storage medium | |
| CN115056829A (en) | Train motion state estimation method for multi-vehicle type continuous learning | |
| Zheng et al. | A novel deep learning architecture and its application in dynamic load monitoring of the vehicle system | |
| CN110658818B (en) | Multi-objective optimization intelligent operation control prediction method for novel power supply train | |
| CN114462700A (en) | Railway track quality evaluation method and device | |
| CN113705891A (en) | Urban business office complex building parking demand prediction method based on MRA-BAS-BP algorithm | |
| CN112308824A (en) | Curve radius classification identification method and device based on track geometric detection data | |
| Wang et al. | Research on parking space prediction based on long short-term memory | |
| CN112183597A (en) | Unbalanced traffic accident severity analysis method based on spectral clustering | |
| Bohan et al. | Short-term traffic flow prediction based on bus floating car |
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 |