Based On The Research Of Improved Wavelet Neural Network In Short-term Traffic Flow Forecasting

In recent years,with the continuous improvement of living standards,the number of people buying cars has increased.At the same time,road traffic has become more and more overburdened,which has led to a series of problems,such as more traffic jams and more traffic incidents.In order to alleviate the problem caused by the increase of traffic volume,the intelligent traffic management system(ITS)has been developed,which is one of the main means of traffic control and traffic guidance.A good intelligent traffic management system can closely monitor the traffic situation on the road,maximize the traffic efficiency as much as possible,so as to alleviate the traffic congestion,improve the ability of passing through the road,reduce the occurrence of traffic accident,reduce energy consumption and reduce environmental pollution caused by increased vehicle traffic.Therefore,it is very meaningful to study short-term traffic flow forecasting method.Due to the time-varying and nonlinear characteristics of short-term traffic flow data,it is difficult to predict the short-term traffic flow data accurately by general forecasting methods.Therefore,according to the nature of short-time traffic flow and combining with the self-adaptability and powerful learning ability of neural network,improved wavelet neural network is used to predict short-time traffic flow in this paper.The traditional Wavelet Neural Network(WNN)uses gradient descent method to train the network.While gradient descent method is sensitive to the initial value of the network parameters and easy to fall into the local optimum,thus affecting the prediction effect of the WNN network.Therefore,a hybrid optimization algorithm(AAFA_PSO)combined with improved artificial fish swarm algorithm and particle swarm optimization algorithm is proposed in this paper.Then,the AAFA_PSO algorithm is introduced into the wavelet neural network to optimize the parameters of the network,an improved wavelet neural network prediction model(AAFA_PSO_WNN)is proposed.In order to test the performance of the improved wavelet neural network model(AAFA_PSO_WNN)proposed in this paper,a comparative verification experiment is carried out.The simulation results show that the AAFA_PSO_WNN model proposed in this paper has a good prediction performance.After pre-processing the short-term traffic data(data repair,noise reduction and normalization processing),C-C method,empirical method,trial-error method and other methods are used to determine the structure of wavelet neural network.Secondly,the AAFA_PSO_WNN prediction model proposed in this paper is used to forecast the short-term traffic flow data.In order to test the performance of AAFA_PSO_WNN prediction model in forecasting the short-term traffic flow data,ACO_WNN(Ant Colony Optimization Optimizes Wavelet Neural Network),GA_WNN(Genetic Algorithm Optimizes Wavelet Neural Network),and GA_BP(Genetic Algorithm Optimized BP Neural Network)are introduced to the simulation experiment.By analyzing the experimental results data,it is proved that the AAFA_PSO_WNN prediction model proposed in this paper has better performance and stability in shortterm traffic prediction.