Short-Term Traffic Flow Prediction Based On Improved Ant Colony Hybrid Wavelet Neural Network

In today’s era of rapid economic growth,people’s material living standards have gradually improved.In terms of transportation,people go to work and travel more inclined to drive by oneself,at the same time,the purchase volume and frequency of use of vehicles also increase,which is likely to cause a burden of traffic congestion.With the development of artificial intelligence,various intelligent systems have also emerged.In order to reduce traffic pressure,the intelligent transportation system is used by relevant departments.The intelligent transportation system can not only bring people traffic information that is beneficial to their own travel,but also help relevant departments to carry out effective traffic guidance and control.The main basis of traffic guidance and control is accurate,efficient and real-time short-term prediction of traffic flow.Among them,the accuracy of short-term traffic flow prediction has played a decisive role.More and more scholars have begun in-depth research and discussion on short-term traffic flow forecasting.This paper focuses on how to improve the accuracy of short-term traffic flow prediction.In order to improve the accuracy of short-term traffic flow prediction,this paper starts from two aspects of its prediction method and prediction model.The main research contents are:(1)Using wavelet neural network prediction model.The traffic flow data sequence is a time series with highly nonlinear features,while the wavelet neural network can show good results in nonlinear time series identification and prediction.After research,it is decided to use wavelet neural network as the short-term traffic flow prediction model.(2)Improve the short-term traffic flow prediction method.In short-term traffic flow prediction,its prediction error has a certain law.In order to be able to extract the prediction error twice,this paper introduces the error compensation method.At the same time,based on a traffic flow data with a highly nonlinear time series,there is still a linear correlation.In order to improve the prediction accuracy,the wavelet neural network is used to predict the nonlinear part of the traffic flow sequence,and the Kalman filter model is introduced to predict the linear part of the traffic flow sequence,so as to establish a hybrid wavelet neural network prediction.(3)improve the ant colony optimization,Optimize the wavelet neural network prediction model.In view of the shortcomings of wavelet neural network in the prediction of sensitivity to weight and wavelet factor,the ant colony optimization is introduced for optimization.In order to improve the accuracy of the ant colony optimization in the optimization process,the influence of the heuristic function and pheromone concentration in the ant colony optimization is improved.A short-term traffic flow prediction based on improved ant colony optimization for wavelet neural network is established.The main innovation of this paper is to combine the improved ant colony optimization and the hybrid wavelet neural network prediction model to establish a short-term traffic flow prediction based on improved ant colony hybrid wavelet neural network.Based on this,the simulation experiments in the MATLAB2016 b environment were carried out on the traffic flow data sets obtained from the Traffic Data Research Laboratory of Duluth University,Minnesota,USA.The simulation results of four kinds of prediction models based on wavelet neural network,hybrid wavelet neural network,wavelet neural network based on improved ant colony optimization and hybrid wavelet neural network based on improved ant colony are compared.The experimental analysis shows that the short-term traffic flow prediction based on the improved ant colony hybrid wavelet neural network is improved in terms of prediction accuracy and fitness compared with the original prediction method,the improvement in accuracy is reflected in the reduction of the prediction bias.the increase in fitness is reflected in the increase in value from 0.97204 to 0.98676.Due to the increase in prediction complexity,the time consumed by the prediction is also slightly increased,but the real-time performance of short-term traffic flow prediction is still satisfied.From the overall results,this paper based on improved ant colony hybrid wavelet neural network short-term traffic flow prediction has a better advantage in forecasting effect.