Research On Short-term Traffic Flow Prediction Model Optimized By Improved Wavelet Neural Network

Intelligent traffic management system（ITS）is an important research topic in the process of smart city construction,and the research on short-term traffic flow prediction is one of the important contents of the research topic.According to the research status at home and abroad,based on improved wavelet neural network,the research on short-term traffic flow prediction is one of the current research hotspots.Wavelet neural network has good self-learning ability and good prediction accuracy of short-term traffic flow.In the paper,improved genetic algorithm（IGA）and improved method evolutionary algorithm（IMEA）are used to optimize wavelet neural network（WNN）,and a combined prediction model is constructed.The main research work of this paper is as follows.This paper selects appropriate traffic flow data for traffic flow data preprocessing.Due to the problems of incomplete and disordered data in the traffic flow data of the observation station,this paper uses the data repair method to preprocess the data,and normalizes the processed data to facilitate the input of WNN model.Then the improved mind evolutionary algorithm is used to optimize the wavelet neural network to improve the prediction accuracy of the prediction model.Firstly,a wavelet neural network short-term traffic flow prediction model（WNN）is constructed,and then the improved genetic algorithm is used to optimize the wavelet neural network to construct the prediction model（IGA-WNN）.At the same time,in view of the high randomness of the convergence and alienation operation of the mind evolutionary algorithm and the insufficient utilization of the announcement information,a method similar to the particle moving and updating position in the particle swarm optimization（PSO）algorithm is introduced to build a wavelet neural network prediction model（IMEA-WNN）based on the improved mind evolutionary algorithm to predict the short-term traffic flow after the convergence operation.The model fitting degree EC_error reaches 93.3%.Finally,a combined short-term traffic flow prediction model is constructed.Based on the IMEA-WNN prediction model and taking the model prediction residual as the data set,the paper uses the LSTM model and error compensation method to predict the residual value respectively,constructs the IMEA-WNN-LSTM combined model and IMEA-WNN-EC combined model,and adds the residual value and the predicted value of IMEA-WNN EC_error of the two combined prediction models are 96.3%and 97.4%respectively.It is proved that the combined model has practical application value.