CN110070713A - A kind of traffic flow forecasting method based on two-way nested-grid ocean LSTM neural network - Google Patents

A kind of traffic flow forecasting method based on two-way nested-grid ocean LSTM neural network Download PDF

Info

Publication number
CN110070713A
CN110070713A CN201910298059.6A CN201910298059A CN110070713A CN 110070713 A CN110070713 A CN 110070713A CN 201910298059 A CN201910298059 A CN 201910298059A CN 110070713 A CN110070713 A CN 110070713A
Authority
CN
China
Prior art keywords
traffic flow
data
road
model
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910298059.6A
Other languages
Chinese (zh)
Other versions
CN110070713B (en
Inventor
徐东伟
彭鹏
王永东
戴宏伟
宣琦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN201910298059.6A priority Critical patent/CN110070713B/en
Publication of CN110070713A publication Critical patent/CN110070713A/en
Application granted granted Critical
Publication of CN110070713B publication Critical patent/CN110070713B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/01—Detecting movement of traffic to be counted or controlled
    • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125—Traffic data processing
    • G08G1/0129—Traffic data processing for creating historical data or processing based on historical data

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Primary Health Care (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Educational Administration (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Traffic Control Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A kind of traffic flow forecasting method based on two-way nested-grid ocean LSTM neural network, this method obtains the traffic flow data in prediction section and K most relevant road segments based on road traffic flow correlation matrix, constructs road traffic flow space-time matrix data set and carries out Data Serialization processing；Then two-way nested-grid ocean LSTM neural network prediction model is constructed, prediction model loss function is defined, combined training collection data complete model training；The input of model after finally using test set data as training, realizes the real-time prediction of test set traffic flow modes and Definition Model evaluation criterion, carries out error analysis.The present invention by improve LSTM unit time level effect and look to the future, historical traffic stream mode and standing state contact, the temporal characteristics extractability of road traffic flow data is improved, to improve the precision of prediction of road traffic flow.

Description

Traffic flow prediction method based on bidirectional nested LSTM neural network
Technical Field
The invention belongs to the field of traffic prediction, and relates to a traffic flow prediction method based on a bidirectional nested LSTM neural network.
Background
With the continuous advance of urban modernization, the living standard of people is continuously improved, the occupied amount of vehicles per capita is continuously increased, the existing urban road traffic network construction cannot meet the increasing road traffic travel requirements, and people begin to construct an intelligent traffic system to relieve the road congestion problem under the limited road traffic construction. The road traffic flow prediction serving as an important component of the intelligent traffic system not only can assist a road traffic management department in controlling and inducing the road traffic flow, but also can provide a basis for making more reasonable travel decisions for people.
The existing road traffic flow prediction method can be mainly divided into three categories: the first type is a traditional prediction model based on mathematical statistics, such as a linear regression model, a Kalman filtering model and the like, the model is convenient to construct, but the nonlinear capturing capability is not strong, the adaptability is poor, and the prediction precision is not high. The second type is a prediction model based on machine learning, such as a support vector machine model, a neural network model and the like, wherein the traditional machine learning prediction model in the model has better nonlinear capturing capability but is not suitable for big data; the deep learning prediction model has good data feature capture capability and is suitable for big data, but has a certain problem of model overfitting and is complex to construct. The third category is a combination model, which has better prediction accuracy but increased complexity compared to a single model.
Disclosure of Invention
In order to overcome the defect of low prediction precision of the conventional traffic flow prediction method, the invention aims to provide a traffic flow prediction method based on a bidirectional nested LSTM neural network. The method fully excavates the time characteristic change in the road traffic state data by increasing the effective time level of the LSTM neural network and modeling by considering the potential influence of history, future traffic flow state and current traffic flow state, thereby further improving the prediction precision of the existing LSTM.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a traffic flow prediction method based on a bidirectional nested LSTM neural network comprises the following steps:
(1) selecting a certain area of a road traffic network as a research object, acquiring road traffic flow data, and performing road section traffic flow correlation analysis to form a road traffic correlation matrix;
(2) acquiring relevant road section traffic flow data for data preprocessing according to the traffic flow correlation analysis result; constructing a road traffic flow space-time matrix data set, and dividing a training set and a test set; serializing the data set;
(3) constructing a bidirectional nested LSTM neural network, taking a traffic flow space-time data training set as model input, defining a prediction model loss function, and completing model training;
(4) inputting a traffic flow space-time data test set as a prediction model to realize the prediction of the future state of the traffic flow; and defining a model evaluation index and carrying out error analysis on a model prediction result.
Further, the process of the step (1) is as follows:
1.1, selecting a certain area of a road traffic network as a research object to obtain road traffic flow data;
1.2 calculating the road traffic flow correlation matrix
Selecting traffic flow data of different road sections in a research area at the same historical stage, and calculating a regional road section correlation matrix based on the traffic flow data; defining the traffic flow data of the road section i in the period a-b as Xi＝{xia,xia+1,…,xibThe traffic flow data of the road section j in the a-b time period is Xj＝{xja,xja+1,…,xjbCalculating a correlation coefficient rho (X) of the sections i and ji,Xj) The mathematical expression is as follows:
wherein,representing the average of traffic flow for segment i over time periods a-b,representing the average value of traffic flow for segment j over time periods a-b,representing the standard deviation of road segment j over the time period a-b,representing the standard deviation of road segment j over the time period a-b,representing the covariance of traffic flows for segments i and j during the a-b period;
then when there are n road segments in the study area, then the road segment traffic flow correlation matrix in the study area is as follows:
still further, the process of step (2) is:
2.1 road traffic flow data preprocessing
Selecting K most relevant road sections of each road section according to the traffic flow correlation matrix and the correlation from large to small;
acquiring traffic flow data of a corresponding road section and the K most relevant road section, and performing data normalization processing, wherein a data normalization mathematical expression is as follows:
wherein x isitIs the original flow data, minx, of the section i at time tiFor the minimum value in the raw flow data for the section i, maxxiIs the maximum value, x, in the raw flow data for the section ii′tThe normalized road section i flow data is obtained;
2.2 construction of a road traffic flow spatio-temporal matrix dataset
Obtaining traffic flow data after normalization of corresponding road sections and K most relevant road sections, and constructing a road traffic flow space-time matrix data set, wherein the construction form of the data set is as follows:
the row vector of the data set represents traffic flow data of different road sections at the same moment, the column vector represents traffic flow data of the same road section at different moments, and m is the data length of each road section in the data set;
selecting a certain proportion to divide the data set into a training set and a test set;
2.3 data serialization
In order to meet the requirement of the input data size of the prediction model, data serialization processing is carried out on the training set and the test set, and the data serialization processing result of the training set is as follows:
wherein s is the size of a time window, p is a prediction step length, q is the data length of a training set, and Train' is the result of the serialization processing of the training set;
and the test set adopts the same serialization method to carry out data serialization.
Further, the step (3) comprises the following steps:
3.1 construction of bidirectional nested LSTM neural network
Firstly, constructing a nested LSTM neural unit, wherein the mathematical expression of the information processing process of the neural unit is as follows:
it＝σ(xtWxi+ht-1Whi+bi)
ft＝σ(xtWxf+ht-1Whf+bf)
ot＝σ(xtWxo+ht-1Who+bo)
ht＝ot⊙σ(ct)
wherein ⊙ represents a dot product, σ (-) represents a sigmoid function, and Wxf、Wxi、WxoInput weight matrix, W, representing external forgetting gate, input gate, output gatehf、Whi、WhoRepresents the outsideOutput weight matrix of forgetting gate, input gate, output gate at previous moment, bf、bi、boA bias matrix representing external forgetting gates, input gates, output gates,an input weight matrix representing an internal forgetting gate, an input gate, a state unit, an output gate, the output weight matrix at the previous moment of representing the internal forgetting gate, the input gate, the state unit and the output gate,a bias matrix representing internal forgetting gates, input gates, state cells, output gates. i.e. it、ft、ct、ot、htShowing the output of the external input gate, the forgetting gate, the unit state, the output gate, and the memory unit,the output of the internal input gate, the forgetting gate, the unit state, the output gate and the memory unit is shown;
then constructing a bidirectional nested LSTM neural network on the basis of the nested LSTM units, wherein M isiTo nest LSTM units, then, the traffic flow prediction process based on the bi-directional nested LSTM neural network is described as follows:
zt＝f(Wxzxt+Whzzt-1)
wherein x istIs the traffic flow state at time t, ztFor the forward output of the hidden layer at time t-1 of the model,for the reverse output of the hidden layer at model time t +1, ytF is a nested LSTM unit function, g is a relu function, W is a forward weight matrix corresponding to each part,the inverse weight matrix corresponding to each part;
3.2 defining the prediction model loss function
In order to enable the predicted value of the model to be closer to the actual value, the invention adopts the mean square error as the loss function of the prediction model, and the mathematical expression of the loss function is as follows:
where j is the number of training samples, yiIs the predicted value of the ith sample, yi' is the actual value of the ith sample;
3.3 using the training set as model input to complete model training
And taking the training set as model input, training the model by adopting an Adam optimizer based on a back propagation algorithm according to a defined model loss function, selecting a model training result under relatively optimal parameters as a final result, and finishing model training.
The process of the step (4) is as follows:
4.1: realizing the prediction of the future state of the traffic flow
Taking the test set as the input of the trained model, and predicting the future state of the traffic flow of the test set;
4.2: defining model evaluation index and carrying out error analysis
Testing the performance of the model by using the absolute mean-square error MAE, the relative percent absolute mean-square error MAPE and the root-mean-square error RMSE, and carrying out error analysis on the prediction result of the test set;
wherein, yi' is traffic flow label data, yiThe predicted value of the traffic flow is obtained.
The invention has the beneficial effects that: the invention provides a traffic flow prediction method based on a bidirectional nested LSTM neural network. Compared with the prior art, the method can deeply mine the nonlinear characteristics and the space-time characteristics in the traffic flow data. In addition, compared with the traditional LSTM traffic flow prediction method, the method has more effective time hierarchy relationship, can process information of longer time scale, and considers the influence of the future state on the current state during model training.
Drawings
FIG. 1 is a schematic diagram of the structure of a nested LSTM neural unit;
FIG. 2 is a schematic diagram of a bidirectional nested LSTM neural network architecture;
FIG. 3 is a comparison of traffic flow predictions and actual values based on a bi-directional nested LSTM neural network model.
Detailed Description
The invention will be further explained with reference to the drawings.
Referring to fig. 1 to 3, a traffic flow prediction method based on a bidirectional nested LSTM neural network includes the following steps:
(1) acquiring a road traffic correlation matrix based on the road traffic flow state correlation, wherein the process is as follows;
1.1, selecting a certain area of a road traffic network as a research object to obtain road traffic flow data;
1.2 calculating the road traffic flow correlation matrix
Selecting traffic flow data of different road sections in the research area in the same historical stage, and calculating a regional road section correlation matrix based on the traffic flow data. Defining the traffic flow data of the road section i in the period a-b as Xi＝{xia,xia+1,…,xibThe traffic flow data of the road section j in the a-b time period is Xj＝{xja,xja+1,…,xjbCalculating a correlation coefficient rho (X) of the sections i and ji,Xj) The mathematical expression is as follows:
wherein,representing the average of traffic flow for segment i over time periods a-b,representing the average value of traffic flow for segment j over time periods a-b,representing the standard deviation of road segment j over the time period a-b,representing the standard deviation of road segment j over the time period a-b,representing the covariance of traffic flows for segments i and j over the period a-b.
Then when there are n road segments in the study area, then the road segment traffic flow correlation matrix in the study area is as follows:
(2) constructing a road traffic flow space-time matrix data set based on data preprocessing and serializing the data; the process is as follows:
2.1 road traffic flow data preprocessing
And selecting K most relevant road sections (the K value can be set according to actual requirements) of each road section according to the traffic flow correlation matrix and the correlation from large to small.
And acquiring traffic flow data of the corresponding road section and the K most relevant road section, and performing data normalization processing. The data normalization mathematical expression is as follows:
wherein x isitIs the original flow data, minx, of the section i at time tiFor the minimum value in the raw flow data for the section i, maxxiIs the maximum value, x, in the raw flow data for the section ii′tThe normalized road section i flow data is obtained.
2.2 construction of a road traffic flow spatio-temporal matrix dataset
And acquiring traffic flow data of the corresponding road section and the K most relevant road section after normalization, and constructing a road traffic flow space-time matrix data set. The dataset construction form is as follows:
the row vector of the data set represents traffic flow data of different road sections at the same time, the column vector represents traffic flow data of the same road section at different times, and m is the data length of each road section in the data set.
And selecting a certain proportion to divide the data set into a training set and a test set.
2.3 data serialization
And in order to meet the requirement of the input data size of the prediction model, performing data serialization processing on the training set and the test set. The training set data serialization processing results are as follows:
wherein s is the size of the time window, p is the prediction step length, q is the data length of the training set, and Train' is the result of the serialization of the training set.
And the test set adopts the same serialization method to carry out data serialization.
(3) Inputting the traffic flow space-time matrix training set into a bidirectional nested LSTM neural network to complete model training; the process is as follows:
3.1 construction of bidirectional nested LSTM neural network
First, a nested LSTM neural unit is constructed, the unit structure of which is shown in fig. 1. The neural unit process information mathematical expression is as follows:
it＝σ(xtWxi+ht-1Whi+bi)
ft＝σ(xtWxf+ht-1Whf+bf)
ot＝σ(xtWxo+ht-1Who+bo)
ht＝ot⊙σ(ct)
wherein ⊙ represents a dot product, σ (-) represents a sigmoid function, and Wxf、Wxi、WxoInput weight matrix, W, representing external forgetting gate, input gate, output gatehf、Whi、WhoRepresenting the output weight matrix of the external forgetting gate, the input gate, the output gate at the previous moment, bf、bi、boA bias matrix representing external forgetting gates, input gates, output gates, an input weight matrix representing an internal forgetting gate, an input gate, a state unit, an output gate,the output weight matrix at the previous moment of representing the internal forgetting gate, the input gate, the state unit and the output gate,a bias matrix representing internal forgetting gates, input gates, state cells, output gates. i.e. it、ft、ct、ot、htShowing the output of the external input gate, the forgetting gate, the unit state, the output gate, and the memory unit,indicating the output of the internal input gate, the forgetting gate, the cell state, the output gate, and the memory cell.
Then, a bidirectional nested LSTM neural network is constructed on the basis of the nested LSTM units, and the schematic diagram of the network structure is shown in FIG. 2, wherein M isiIs a nested LSTM cell. Then, the traffic flow prediction process based on the bidirectional nested LSTM neural network may be described as the following process:
zt＝f(Wxzxt+Whzzt-1)
wherein x istIs the traffic flow state at time t, ztFor the forward output of the hidden layer at time t-1 of the model,for the reverse output of the hidden layer at model time t +1, ytF is a nested LSTM unit function, g is a relu function, W is a forward weight matrix corresponding to each part,and the inverse weight matrix corresponds to each part.
3.2 defining the prediction model loss function
In order to enable the predicted value of the model to be closer to the actual value, the invention adopts the mean square error as the loss function of the prediction model, and the mathematical expression of the loss function is as follows:
where j is the number of training samples, yiIs the predicted value of the ith sample, yi' is the actual value of the ith sample.
3.3 using the training set as model input to complete model training
And taking the training set as model input, and training the model by adopting an Adam optimizer based on a back propagation algorithm according to a defined model loss function. And selecting a model training result under the relatively optimal parameters as a final result to finish the model training.
(4) Predicting the traffic flow state and carrying out error analysis, wherein the process comprises the following steps:
4.1: realizing the prediction of the future state of the traffic flow
And (5) taking the test set as the input of the trained model, and predicting the future state of the traffic flow of the test set.
4.2: defining model evaluation index and carrying out error analysis
And (4) checking the performance of the model by using the absolute mean-square error MAE, the relative percent absolute mean-square error MAPE and the root-mean-square error RMSE, and carrying out error analysis on the prediction result of the test set.
Wherein, yi' is traffic flow label data, yiThe predicted value of the traffic flow is obtained.
According to the data in the practical experiment, the implementation process is as follows:
1) selecting experimental data
The original traffic flow data set comprises 29 continuous days of traffic flow data of two ring sections in Beijing city, and the data sampling interval T is 2 min.
In the invention, the data of the previous 25 days is taken as a training set to train the prediction model. The rest data is a test set, and the effectiveness of the method is verified.
2) Parameter determination
The main parameters related to the invention comprise the most relevant neighbor number k, the time window size s, the prediction step length p,
Number of hidden layer units, hidden _ num, and number of output layer units, output _ num. Through experimental comparison, the final parameters are determined as k ═ 3, s ═ 5, p ═ 1, hidden _ num ═ 64, and output _ num ═ 1.
3) Results of the experiment
In order to test the feasibility and the effectiveness of the invention, the invention takes the HI2075a road section as an example to carry out prediction precision test, and the measurement indexes are mean square error of absolute value (MAE), relative absolute mean square error of percentage (MAPE) and Root Mean Square Error (RMSE). The actual prediction results are shown in fig. 3, and the evaluation analysis of the actual prediction results is shown in table 1.
MAE MAPE(％) RMSE
12.97 9.13 168.24
Table 1.

Claims (5)

1. A traffic flow prediction method based on a bidirectional nested LSTM neural network is characterized by comprising the following steps:
(1) selecting a certain area of a road traffic network as a research object, acquiring road traffic flow data, and performing road section traffic flow correlation analysis to form a road traffic correlation matrix;
(2) acquiring relevant road section traffic flow data for data preprocessing according to the traffic flow correlation analysis result; constructing a road traffic flow space-time matrix data set, and dividing a training set and a test set; serializing the data set;
(3) constructing a bidirectional nested LSTM neural network, taking a traffic flow space-time data training set as model input, defining a prediction model loss function, and completing model training;
(4) inputting a traffic flow space-time data test set as a prediction model to realize the prediction of the future state of the traffic flow; and defining a model evaluation index and carrying out error analysis on a model prediction result.
2. The traffic flow prediction method based on the bidirectional nested LSTM neural network as claimed in claim 1, wherein the process of step (1) is:
1.1, selecting a certain area of a road traffic network as a research object to obtain road traffic flow data;
1.2 calculating the road traffic flow correlation matrix
Selecting traffic flow data of different road sections in a research area at the same historical stage, and calculating a regional road section correlation matrix based on the traffic flow data; defining the traffic flow data of the road section i in the period a-b as Xi＝{xia,xia+1,…,xibThe traffic flow data of the road section j in the a-b time period is Xj＝{xja,xja+1,…,xjbCalculating a correlation coefficient rho (X) of the sections i and ji,Xj) The mathematical expression is as follows:
wherein,representing the average of traffic flow for segment i over time periods a-b,representing the average value of traffic flow for segment j over time periods a-b,representing the standard deviation of road segment j over the time period a-b,representing the standard deviation of road segment j over the time period a-b,representing the covariance of traffic flows for segments i and j during the a-b period;
then when there are n road segments in the study area, then the road segment traffic flow correlation matrix in the study area is as follows:
3. a traffic flow prediction method based on bidirectional nested LSTM neural network as claimed in claim 1 or 2, wherein the process of step (2) is:
2.1 road traffic flow data preprocessing
Selecting K most relevant road sections of each road section according to the traffic flow correlation matrix and the correlation from large to small;
acquiring traffic flow data of a corresponding road section and the K most relevant road section, and performing data normalization processing, wherein a data normalization mathematical expression is as follows:
wherein x isitIs the original flow data, minx, of the section i at time tiFor the minimum value in the raw flow data for the section i, maxxiIs the maximum value, x 'in the segment i raw flow data'itThe normalized road section i flow data is obtained;
2.2 construction of a road traffic flow spatio-temporal matrix dataset
Obtaining traffic flow data after normalization of corresponding road sections and K most relevant road sections, and constructing a road traffic flow space-time matrix data set, wherein the construction form of the data set is as follows:
the row vector of the data set represents traffic flow data of different road sections at the same moment, the column vector represents traffic flow data of the same road section at different moments, and m is the data length of each road section in the data set;
selecting a certain proportion to divide the data set into a training set and a test set;
2.3 data serialization
In order to meet the requirement of the input data size of the prediction model, data serialization processing is carried out on the training set and the test set, and the data serialization processing result of the training set is as follows:
wherein s is the size of a time window, p is a prediction step length, q is the data length of a training set, and Train' is the result of the serialization processing of the training set;
and the test set adopts the same serialization method to carry out data serialization.
4. A traffic flow prediction method based on a bidirectional nested LSTM neural network as claimed in claim 1 or 2, wherein the process of step (3) is:
3.1 construction of bidirectional nested LSTM neural network
Firstly, constructing a nested LSTM neural unit, wherein the mathematical expression of the information processing process of the neural unit is as follows:
it＝σ(xtWxi+ht-1Whi+bi)
ft＝σ(xtWxf+ht-1Whf+bf)
ot＝σ(xtWxo+ht-1Who+bo)
ht＝ot⊙σ(ct)
wherein ⊙ represents a dot product, σ (-) represents a sigmoid function, and Wxf、Wxi、WxoInput weight matrix, W, representing external forgetting gate, input gate, output gatehf、Whi、WhoRepresenting the output weight matrix of the external forgetting gate, the input gate, the output gate at the previous moment, bf、bi、boA bias matrix representing external forgetting gates, input gates, output gates,an input weight matrix representing an internal forgetting gate, an input gate, a state unit, an output gate, the output weight matrix at the previous moment of representing the internal forgetting gate, the input gate, the state unit and the output gate,bias matrix representing internal forgetting gate, input gate, status cell, output gate, it、ft、ct、ot、htShowing the output of the external input gate, the forgetting gate, the unit state, the output gate, and the memory unit,the output of the internal input gate, the forgetting gate, the unit state, the output gate and the memory unit is shown;
then constructing a bidirectional nested LSTM neural network on the basis of the nested LSTM units, wherein M isiTo nest LSTM units, then, the traffic flow prediction process based on the bi-directional nested LSTM neural network is described as follows:
zt＝f(Wxzxt+Whzzt-1)
wherein x istIs the traffic flow state at time t, ztFor the forward output of the hidden layer at time t-1 of the model,for the reverse output of the hidden layer at model time t +1, ytF is a nested LSTM unit function, g is a relu function, W is a forward weight matrix corresponding to each part,the inverse weight matrix corresponding to each part;
3.2 defining the prediction model loss function
In order to enable the predicted value of the model to be closer to the actual value, the invention adopts the mean square error as the loss function of the prediction model, and the mathematical expression of the loss function is as follows:
where j is the number of training samples, yiIs a predicted value of the ith sample, y'iIs the actual value of the ith sample;
3.3 using the training set as model input to complete model training
And taking the training set as model input, training the model by adopting an Adam optimizer based on a back propagation algorithm according to a defined model loss function, selecting a model training result under relatively optimal parameters as a final result, and finishing model training.
5. The traffic flow prediction method based on the bidirectional nested LSTM neural network as claimed in claim 1 or 2, wherein the process of step (4) is:
4.1: realizing the prediction of the future state of the traffic flow
Taking the test set as the input of the trained model, and predicting the future state of the traffic flow of the test set;
4.2: defining model evaluation index and carrying out error analysis
Testing the performance of the model by using the absolute mean-square error MAE, the relative percent absolute mean-square error MAPE and the root-mean-square error RMSE, and carrying out error analysis on the prediction result of the test set;
wherein, y'iFor traffic flow label data, yiThe predicted value of the traffic flow is obtained.
CN201910298059.6A 2019-04-15 2019-04-15 Traffic flow prediction method based on bidirectional nested LSTM neural network Active CN110070713B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910298059.6A CN110070713B (en) 2019-04-15 2019-04-15 Traffic flow prediction method based on bidirectional nested LSTM neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910298059.6A CN110070713B (en) 2019-04-15 2019-04-15 Traffic flow prediction method based on bidirectional nested LSTM neural network

Publications (2)

Publication Number Publication Date
CN110070713A true CN110070713A (en) 2019-07-30
CN110070713B CN110070713B (en) 2021-01-01

Family

ID=67367668

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910298059.6A Active CN110070713B (en) 2019-04-15 2019-04-15 Traffic flow prediction method based on bidirectional nested LSTM neural network

Country Status (1)

Country Link
CN (1) CN110070713B (en)

Cited By (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110569591A (en) * 2019-09-03 2019-12-13 西北工业大学 agile development demand change prediction method based on Bi-LSTM
CN110837888A (en) * 2019-11-13 2020-02-25 大连理工大学 Traffic missing data completion method based on bidirectional cyclic neural network
CN110889347A (en) * 2019-11-15 2020-03-17 山东大学 Density traffic flow counting method and system based on space-time counting characteristics
CN111241243A (en) * 2020-01-13 2020-06-05 华中师范大学 Knowledge measurement-oriented test question, knowledge and capability tensor construction and labeling method
CN111241744A (en) * 2020-01-07 2020-06-05 浙江大学 Low-pressure casting machine time sequence data abnormity detection method based on bidirectional LSTM
CN111275971A (en) * 2020-02-18 2020-06-12 山西交通控股集团有限公司 Expressway traffic flow prediction method
CN111291924A (en) * 2020-01-17 2020-06-16 同济大学 Geometric algebraic deep neural network model method for long-term traffic speed prediction
CN111598325A (en) * 2020-05-11 2020-08-28 浙江工业大学 Traffic speed prediction method based on hierarchical clustering and hierarchical attention mechanism
CN111709549A (en) * 2020-04-30 2020-09-25 东华大学 Navigation reminding method for short-time traffic flow prediction based on SVD-PSO-LSTM
CN111833173A (en) * 2020-06-03 2020-10-27 百维金科(上海)信息科技有限公司 LSTM-based third-party platform payment fraud online detection method
CN111895923A (en) * 2020-07-07 2020-11-06 上海辰慧源科技发展有限公司 Method for fitting and measuring thickness of thin film
CN112070280A (en) * 2020-08-19 2020-12-11 贵州民族大学 Real-time traffic flow parallel prediction method, system, terminal and storage medium
CN112270355A (en) * 2020-10-28 2021-01-26 长沙理工大学 Active safety prediction method based on big data technology and SAE-GRU
CN112382089A (en) * 2020-11-11 2021-02-19 湖南大学 Traffic junction node flow prediction method based on road network directed graph and parallel long-time memory network
CN112446516A (en) * 2019-08-27 2021-03-05 北京理工大学 Travel prediction method and device
CN112529299A (en) * 2020-12-11 2021-03-19 东南大学 Short traffic flow prediction method based on ARIMA and LSTM mixed neural network
CN112561146A (en) * 2020-12-08 2021-03-26 哈尔滨工程大学 Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM
CN112561174A (en) * 2020-12-18 2021-03-26 西南交通大学 Method for predicting geothermal energy production based on LSTM and MLP superimposed neural network
CN112766603A (en) * 2021-02-01 2021-05-07 湖南大学 Traffic flow prediction method, system, computer device and storage medium
CN112907969A (en) * 2021-02-02 2021-06-04 中国科学院计算技术研究所 Method and system for predicting road traffic flow
CN112906945A (en) * 2021-01-27 2021-06-04 昭通亮风台信息科技有限公司 Traffic flow prediction method, system and computer readable storage medium
CN112927501A (en) * 2021-01-21 2021-06-08 南京理工大学 Urban road network space-time traffic state collaborative prediction method based on dynamic factor model
CN112946187A (en) * 2021-01-22 2021-06-11 西安科技大学 Refuge chamber real-time state monitoring method based on neural network
CN112967498A (en) * 2021-02-02 2021-06-15 重庆首讯科技股份有限公司 Short-term traffic flow prediction method based on dynamic space-time correlation characteristic optimization
CN112989548A (en) * 2019-12-17 2021-06-18 南京理工大学 Urban road traffic flow simulation deduction method based on multi-model combination
CN112990598A (en) * 2021-03-31 2021-06-18 浙江禹贡信息科技有限公司 Reservoir water level time sequence prediction method and system
CN113129585A (en) * 2021-03-05 2021-07-16 浙江工业大学 Road traffic flow prediction method based on graph aggregation mechanism of reconstructed traffic network
CN113408781A (en) * 2021-04-30 2021-09-17 南通大学 Encoder-Decoder-based long-term traffic flow prediction method
CN113420414A (en) * 2021-05-27 2021-09-21 四川大学 Short-term traffic flow prediction model based on dynamic space-time analysis
CN113435124A (en) * 2021-06-29 2021-09-24 北京工业大学 Water quality space-time correlation prediction method based on long-time and short-time memory and radial basis function neural network
CN113487855A (en) * 2021-05-25 2021-10-08 浙江工业大学 Traffic flow prediction method based on EMD-GAN neural network structure
CN114255591A (en) * 2021-12-17 2022-03-29 重庆中信科信息技术有限公司 Short-term traffic flow prediction method and device considering space-time correlation and storage medium
CN114283584A (en) * 2021-12-31 2022-04-05 云控智行(上海)汽车科技有限公司 Expressway road condition prediction method under intelligent network connection environment and computer readable storage medium
CN114495507A (en) * 2022-02-25 2022-05-13 南京工业大学 Traffic flow prediction method integrating space-time attention neural network and traffic model
CN114944057A (en) * 2022-04-21 2022-08-26 中山大学 Road network traffic flow data restoration method and system
CN115171372A (en) * 2022-06-20 2022-10-11 青岛海信网络科技股份有限公司 Traffic anomaly detection method, equipment and device
CN115394084A (en) * 2022-08-29 2022-11-25 郑州轻工业大学 NMF-BilSTM-based urban road network short-term traffic flow prediction method
CN116913098A (en) * 2023-09-14 2023-10-20 华东交通大学 Short-time traffic flow prediction method integrating air quality and vehicle flow data

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106530687A (en) * 2016-10-13 2017-03-22 北京交通大学 Time-space-attribute-based method for measuring traffic network node importance degrees
CN107154150A (en) * 2017-07-25 2017-09-12 北京航空航天大学 A kind of traffic flow forecasting method clustered based on road with double-layer double-direction LSTM
CN108510741A (en) * 2018-05-24 2018-09-07 浙江工业大学 A kind of traffic flow forecasting method based on Conv1D-LSTM neural network structures
CN109243172A (en) * 2018-07-25 2019-01-18 华南理工大学 Traffic flow forecasting method based on genetic algorithm optimization LSTM neural network
CN109410575A (en) * 2018-10-29 2019-03-01 北京航空航天大学 A kind of road network trend prediction method based on capsule network and the long Memory Neural Networks in short-term of nested type

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106530687A (en) * 2016-10-13 2017-03-22 北京交通大学 Time-space-attribute-based method for measuring traffic network node importance degrees
CN107154150A (en) * 2017-07-25 2017-09-12 北京航空航天大学 A kind of traffic flow forecasting method clustered based on road with double-layer double-direction LSTM
CN108510741A (en) * 2018-05-24 2018-09-07 浙江工业大学 A kind of traffic flow forecasting method based on Conv1D-LSTM neural network structures
CN109243172A (en) * 2018-07-25 2019-01-18 华南理工大学 Traffic flow forecasting method based on genetic algorithm optimization LSTM neural network
CN109410575A (en) * 2018-10-29 2019-03-01 北京航空航天大学 A kind of road network trend prediction method based on capsule network and the long Memory Neural Networks in short-term of nested type

Cited By (56)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112446516A (en) * 2019-08-27 2021-03-05 北京理工大学 Travel prediction method and device
CN110569591A (en) * 2019-09-03 2019-12-13 西北工业大学 agile development demand change prediction method based on Bi-LSTM
CN110837888A (en) * 2019-11-13 2020-02-25 大连理工大学 Traffic missing data completion method based on bidirectional cyclic neural network
CN110889347A (en) * 2019-11-15 2020-03-17 山东大学 Density traffic flow counting method and system based on space-time counting characteristics
CN112989548A (en) * 2019-12-17 2021-06-18 南京理工大学 Urban road traffic flow simulation deduction method based on multi-model combination
CN111241744B (en) * 2020-01-07 2022-08-09 浙江大学 Low-pressure casting machine time sequence data abnormity detection method based on bidirectional LSTM
CN111241744A (en) * 2020-01-07 2020-06-05 浙江大学 Low-pressure casting machine time sequence data abnormity detection method based on bidirectional LSTM
CN111241243B (en) * 2020-01-13 2023-05-26 华中师范大学 Test question, knowledge and capability tensor construction and labeling method oriented to knowledge measurement
CN111241243A (en) * 2020-01-13 2020-06-05 华中师范大学 Knowledge measurement-oriented test question, knowledge and capability tensor construction and labeling method
CN111291924A (en) * 2020-01-17 2020-06-16 同济大学 Geometric algebraic deep neural network model method for long-term traffic speed prediction
CN111275971A (en) * 2020-02-18 2020-06-12 山西交通控股集团有限公司 Expressway traffic flow prediction method
CN111709549A (en) * 2020-04-30 2020-09-25 东华大学 Navigation reminding method for short-time traffic flow prediction based on SVD-PSO-LSTM
CN111709549B (en) * 2020-04-30 2022-10-21 东华大学 SVD-PSO-LSTM-based short-term traffic flow prediction navigation reminding method
CN111598325A (en) * 2020-05-11 2020-08-28 浙江工业大学 Traffic speed prediction method based on hierarchical clustering and hierarchical attention mechanism
CN111833173A (en) * 2020-06-03 2020-10-27 百维金科(上海)信息科技有限公司 LSTM-based third-party platform payment fraud online detection method
CN111895923A (en) * 2020-07-07 2020-11-06 上海辰慧源科技发展有限公司 Method for fitting and measuring thickness of thin film
CN112070280A (en) * 2020-08-19 2020-12-11 贵州民族大学 Real-time traffic flow parallel prediction method, system, terminal and storage medium
CN112070280B (en) * 2020-08-19 2023-10-31 贵州民族大学 Real-time traffic flow parallel prediction method, system, terminal and storage medium
CN112270355A (en) * 2020-10-28 2021-01-26 长沙理工大学 Active safety prediction method based on big data technology and SAE-GRU
CN112270355B (en) * 2020-10-28 2023-12-05 长沙理工大学 Active safety prediction method based on big data technology and SAE-GRU
CN112382089A (en) * 2020-11-11 2021-02-19 湖南大学 Traffic junction node flow prediction method based on road network directed graph and parallel long-time memory network
CN112561146B (en) * 2020-12-08 2023-04-18 哈尔滨工程大学 Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM
CN112561146A (en) * 2020-12-08 2021-03-26 哈尔滨工程大学 Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM
CN112529299A (en) * 2020-12-11 2021-03-19 东南大学 Short traffic flow prediction method based on ARIMA and LSTM mixed neural network
CN112561174A (en) * 2020-12-18 2021-03-26 西南交通大学 Method for predicting geothermal energy production based on LSTM and MLP superimposed neural network
CN112927501A (en) * 2021-01-21 2021-06-08 南京理工大学 Urban road network space-time traffic state collaborative prediction method based on dynamic factor model
CN112927501B (en) * 2021-01-21 2022-05-27 南京理工大学 Urban road network space-time traffic state collaborative prediction method based on dynamic factor model
CN112946187A (en) * 2021-01-22 2021-06-11 西安科技大学 Refuge chamber real-time state monitoring method based on neural network
CN112946187B (en) * 2021-01-22 2023-04-07 西安科技大学 Refuge chamber real-time state monitoring method based on neural network
CN112906945A (en) * 2021-01-27 2021-06-04 昭通亮风台信息科技有限公司 Traffic flow prediction method, system and computer readable storage medium
CN112766603A (en) * 2021-02-01 2021-05-07 湖南大学 Traffic flow prediction method, system, computer device and storage medium
CN112907969B (en) * 2021-02-02 2022-04-22 中国科学院计算技术研究所 Method and system for predicting road traffic flow
CN112967498B (en) * 2021-02-02 2022-05-03 重庆首讯科技股份有限公司 Short-term traffic flow prediction method based on dynamic space-time correlation characteristic optimization
CN112967498A (en) * 2021-02-02 2021-06-15 重庆首讯科技股份有限公司 Short-term traffic flow prediction method based on dynamic space-time correlation characteristic optimization
CN112907969A (en) * 2021-02-02 2021-06-04 中国科学院计算技术研究所 Method and system for predicting road traffic flow
CN113129585B (en) * 2021-03-05 2022-03-01 浙江工业大学 Road traffic flow prediction method based on graph aggregation mechanism of reconstructed traffic network
CN113129585A (en) * 2021-03-05 2021-07-16 浙江工业大学 Road traffic flow prediction method based on graph aggregation mechanism of reconstructed traffic network
CN112990598A (en) * 2021-03-31 2021-06-18 浙江禹贡信息科技有限公司 Reservoir water level time sequence prediction method and system
CN113408781B (en) * 2021-04-30 2024-08-06 南通大学 Encoder-Decoder-based long-term traffic flow prediction method
CN113408781A (en) * 2021-04-30 2021-09-17 南通大学 Encoder-Decoder-based long-term traffic flow prediction method
CN113487855A (en) * 2021-05-25 2021-10-08 浙江工业大学 Traffic flow prediction method based on EMD-GAN neural network structure
CN113487855B (en) * 2021-05-25 2022-12-20 浙江工业大学 Traffic flow prediction method based on EMD-GAN neural network structure
CN113420414A (en) * 2021-05-27 2021-09-21 四川大学 Short-term traffic flow prediction model based on dynamic space-time analysis
CN113420414B (en) * 2021-05-27 2022-08-30 四川大学 Short-term traffic flow prediction model based on dynamic space-time analysis
CN113435124A (en) * 2021-06-29 2021-09-24 北京工业大学 Water quality space-time correlation prediction method based on long-time and short-time memory and radial basis function neural network
CN114255591A (en) * 2021-12-17 2022-03-29 重庆中信科信息技术有限公司 Short-term traffic flow prediction method and device considering space-time correlation and storage medium
CN114283584A (en) * 2021-12-31 2022-04-05 云控智行(上海)汽车科技有限公司 Expressway road condition prediction method under intelligent network connection environment and computer readable storage medium
CN114495507A (en) * 2022-02-25 2022-05-13 南京工业大学 Traffic flow prediction method integrating space-time attention neural network and traffic model
CN114944057A (en) * 2022-04-21 2022-08-26 中山大学 Road network traffic flow data restoration method and system
CN114944057B (en) * 2022-04-21 2023-07-25 中山大学 Road network traffic flow data restoration method and system
CN115171372B (en) * 2022-06-20 2023-10-24 青岛海信网络科技股份有限公司 Traffic abnormality detection method, equipment and device
CN115171372A (en) * 2022-06-20 2022-10-11 青岛海信网络科技股份有限公司 Traffic anomaly detection method, equipment and device
CN115394084B (en) * 2022-08-29 2023-07-25 郑州轻工业大学 Urban road network short-time traffic flow prediction method based on NMF-BiLSTM
CN115394084A (en) * 2022-08-29 2022-11-25 郑州轻工业大学 NMF-BilSTM-based urban road network short-term traffic flow prediction method
CN116913098A (en) * 2023-09-14 2023-10-20 华东交通大学 Short-time traffic flow prediction method integrating air quality and vehicle flow data
CN116913098B (en) * 2023-09-14 2023-12-22 华东交通大学 Short-time traffic flow prediction method integrating air quality and vehicle flow data

Also Published As

Publication number Publication date
CN110070713B (en) 2021-01-01

Similar Documents

Publication Publication Date Title
CN110070713B (en) Traffic flow prediction method based on bidirectional nested LSTM neural network
Bi et al. Daily tourism volume forecasting for tourist attractions
CN110570651B (en) Road network traffic situation prediction method and system based on deep learning
Wang et al. Long-term traffic prediction based on lstm encoder-decoder architecture
CN109117987B (en) Personalized traffic accident risk prediction recommendation method based on deep learning
CN110766942B (en) Traffic network congestion prediction method based on convolution long-term and short-term memory network
CN110390349A (en) Bus passenger flow volume based on XGBoost model predicts modeling method
CN113362598B (en) Traffic flow prediction method for expressway service area
CN109034448A (en) Trajectory predictions method based on track of vehicle semantic analysis and deepness belief network
CN109902880A (en) A kind of city stream of people's prediction technique generating confrontation network based on Seq2Seq
CN111144281B (en) Urban rail transit OD passenger flow estimation method based on machine learning
CN109215380B (en) Effective parking space prediction method
CN112863182B (en) Cross-modal data prediction method based on transfer learning
CN112949828A (en) Graph convolution neural network traffic prediction method and system based on graph learning
CN111160622A (en) Scenic spot passenger flow prediction method and device based on hybrid neural network model
CN112598165B (en) Urban functional area transfer flow prediction method and device based on private car data
CN107704970A (en) A kind of Demand-side load forecasting method based on Spark
CN115376317B (en) Traffic flow prediction method based on dynamic graph convolution and time sequence convolution network
CN114565187B (en) Traffic network data prediction method based on graph space-time self-coding network
CN112232483A (en) Flight average fare prediction method combining CNN and LSTM
CN114596726A (en) Parking position prediction method based on interpretable space-time attention mechanism
CN118297775B (en) Urban planning management and control system based on digital twin technology
CN108073978A (en) A kind of constructive method of the ultra-deep learning model of artificial intelligence
CN116933946A (en) Rail transit OD passenger flow prediction method and system based on passenger flow destination structure
CN113688200B (en) Decision tree-based special population action track collection method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant