Vehicle flow predicting method based on integrated LSTM neural network

Abstract

The invention relates to a vehicle flow predicting method based on an integrated LSTM neural network. On the basis of historical data obtained by vehicle flow detection, an integrated LSTM neural network vehicle flow prediction model is established to carry out vehicle flow prediction, so that the generalization error of the prediction model is reduced and the accuracy is improved. The method comprises the following steps that: data preprocessing is carried out; according to a preprocessed vehicle flow time sequence value, a vehicle flow matrix data set is constructed and the vehicle flow of an (n+1)th period of time is predicted by using first n periods of time, wherein each period of time is delta t expressing the time length and the unit is min; a plurality of different LSTM neural network models are constructed by using different initial weights; on the basis of a bagging integrated learning method, a training set and a verification set are constructed; a plurality of LSTM neural networks are trained to obtain an optimized module; a weighting coefficient of the single LSTM model is calculated by using the verification set; and inverse transformation and reverse normalization are carried out on a predicted vehicle flow value to obtain a predicted vehicle flow and integrated weighting is carried out to obtain a vehicle flow value predicted finally by the model.

Description

technical field [0001] The invention belongs to the field of traffic flow forecasting, and relates to a traffic flow forecasting method based on an integrated long-short-term memory (Long Short-Term Memory, LSTM) neural network. The invention belongs to the technical field of intelligent transportation. Background technique [0002] Traffic flow prediction is an important research content of intelligent transportation system. Intelligent Transportation System (Intelligence Transportation System, ITS), also known as intelligent transportation system, is a comprehensive application of science and technology to the entire transportation management system, thereby establishing a large-scale, all-round function, real-time and accurate , Efficient integrated transportation and management system. As an important link in the intelligent transportation system, traffic flow forecasting can realize real-time and dynamic forecasting of traffic flow. The intelligent transportation syst...

AI Technical Summary

Problems solved by technology

Among them, the historical average method has a simple algorithm, but it cannot study the dynamics, uncertainty and nonlinear characteristics of traffic flow; the Kalman filter method has a high accuracy in predicting traffic flow, but is limited by the linear estimation model and cannot adapt to the nonlinear change of traffic flow ; The non-parametric regression method is suitable for nonlinear dynamic systems and conforms to the nonlinear characteristics of traffic flow. It requires a large amount of historical data and establishes internal connections in the sequence to predict traffic flow information after the current time. However, the prediction speed is slow and the parameters Adjustment requires trial and error; among many traffic flow prediction methods, neural networks have received more and more attention due to their flexible model structure, powerful learning and generalization capabilities

However, traffic flow has complex historical dependencies. The traffic flow state at this moment has a certain degree of correlation with the historical traffic flow state at the previous moment, and may lead to changes in the traffic flow state at the next moment.

The traditional feed-forward neural network has no concept of time series and cannot remember the early historical input information, so it is difficult to simulate the dynamics of traffic flow

Moreover, when using neural network algorithms to predict traffic flow, the selection of parameters such as initial weights and the selection of network training sample sets will affect the convergence speed of network gradient descent and the probability of gradient descent to the minimum training error, which often requires technical personnel. A lot of parameter adjustment work has been done based on experience, so the single neural network model has problems such as insufficient accuracy and stability

Images