Short-Term Traffic Flow Prediction Of Wavelet Neural Network Based On Flower Pollination Algorithm

In recent years,with the rapid development of the economy and the growing needs of the people,vehicles have become more and more essential in life.Traffic congestion has gradually become an urgent problem in life,establishing a real-time and efficient intelligent traffic management system become the basic way to meet travel needs.Traffic flow forecasting is a key step in the process of building an intelligent traffic management system.Because traffic flow data has chaotic characteristics,strong interference,random volatility and high complexity,many people have used group intelligence algorithms and wavelet neural networks(Wavelet Neural Network,WNN)to study traffic flow prediction,but the current research results can not to realize the accurate prediction of short-term traffic flow.The flower pollination algorithm(Flower Pollination Algorithm,FPA)is a new group intelligent optimization algorithm proposed in recent years.It has good optimization ability and wide application.Therefore,this paper attempts to use the improved flower pollination algorithm to predict the short-term traffic flow of wavelet neural network.The traffic data in this thesis was obtained from the University of Minnesota Traffic Data Research Center and performs corresponding noise reduction and reconstruction processing on the data.The wavelet neural network is used to predict the traffic flow data.The number of hidden layer nodes in the network structure is determined by trial-and-error method,and the training effect of different wavelet functions is compared by simulation experiments.Finally,the Gaussian first-order partial derivative function is selected as the excitation function.The simulation results show that the WNN used in this thesis can make a more accurate prediction of the overall trend of traffic flow.Then,try to combine the flower pollination algorithm with the WNN to predict the shortterm traffic flow.The flower pollination algorithm is used to optimize the initial parameters of the network structure,which effectively solves the problem that the wavelet neural network requires high initial values of the model parameters.The FPA-WNN short-term traffic flow prediction model is constructed to simulate the experiment.The results show that the accuracy of short-term traffic flow prediction is effectively improved,and the accuracy is improved by 1.379%.In order to further improve the prediction accuracy of short-term traffic flow,this paper proposes an improved flower pollination algorithm,add local and global mutation strategies to the flower pollination algorithm to increase the diversity of the population and introduce a dynamic conversion probability strategy to controlling the conversion between local search and global search in the flower pollination algorithm.Compared with the fixed conversion probability value,the dynamic probability value can find the optimal solution better,thereby improving the prediction accuracy.The IFPA-WNN short-term traffic flow prediction model is established by combining IFPA with WNN.The simulation results show that the IFPAWNN model has better prediction effect,which reduces the prediction deviation of traffic flow peak and valley,and the prediction accuracy reaches 98.104%.Compared with the FPA-WNN prediction model,the accuracy is improved by 0.22%.