Research On Algorithm For Traffic Flow Forecasting For Expressway

Traffic is an essential part of modern life and also an important foundation for the country's economic development.However,the rapid development of the national economy has also posed a serious problem of traffic congestion.In order to alleviate traffic congestion,we devote ourselves to researching the methods of traffic flow forecasting to obtain real-time and accurate forecasting data,which provides guiding suggestions for people's travel decisions and route planning so as to dredge road vehicles and relieve traffic pressure.The research object of this paper is traffic flow data.The purpose of this study is to obtain more accurate traffic flow forecasting result.The main research work of this paper is as follows:(1)This paper first introduces the basic concepts of traffic flow,including the basic parameters of traffic flow,basic characteristics,the process of forecasting and the evaluation index of forecasting.Then,the steps of traffic flow preprocessing are introduced,and the de-noising method based on wavelet soft threshold is emphatically described,the simulation results show that this method has good performance of de-noising in the process of traffic flow data.(2)This paper studies and proposes a traffic flow prediction method based on wavelet neural network(WNN)and autoregressive-integrated moving average(ARIMA)combination model.Firstly,the WNN algorithm with good non-linear fitting ability and the traffic flow prediction model based on WNN model are introduced.Then the ARIMA algorithm with good linear approximation ability and the traffic flow prediction model based on ARIMA model are introduced.Aiming at the fact that the traffic flow data contains both linear and nonlinear rules,this paper proposes a WNN and ARIMA combined model algorithm and the traffic flow prediction model based on the WNN and ARIMA combined model.The simulation results show that the prediction accuracy of the combined model is higher than that of the two single model.(3)This paper studies and proposes a traffic flow prediction method based on Elman neural network model optimized by ant colony algorithm.Firstly,the Elman neural network algorithm with some memory ability and the traffic flow prediction model based on Elman neural network are introduced.Then,the ant colony algorithm with global optimization ability is introduced.Aiming at the shortcoming that the random initialization of Elman neural network weights easily leads to local optimal solution,this paper proposes an algorithm based on Elman neural network optimized by ant colony algorithm and the traffic flow prediction model based on Elman neural network model optimized by ant colony algorithm.The simulation results show that the prediction accuracy of Elman neural network model optimized by ant colony algorithm is better than that of Elman neural network model.(4)This paper studies and proposes a traffic flow prediction algorithm based on individuality factor and long short-term memory network(LSTM).Firstly,the structure and algorithm of LSTM based on recurrent neural network are introduced in detail.Then taking the pile number as the individuality factor,and combining historical traffic flow data as the training samples of the LSTM model,this paper proposes a traffic flow prediction algorithm based on the individuality factor and LSTM.The simulation results show that this algorithm can obtain a good prediction effect.