Investigation on the Traffic Flow Based on Wireless Sensor Network Technologies Combined with FA-BPNN Models

Abstract

In this paper, both the four kinds BPNN models can predict the trends of Lozi and Tent chaotic time series. In terms of the predicted effect, the traditional BPNN model performs the worst effect because its predicted effect is far from the real data, while the improved GA-BPNN model and PSO-BPNN model reflect basically the real data. Among these, the proposed FA-BPNN model is the best, which predicted results are basically coincident with the real data. In the respect of the running time, the traditional BPNN model has the longest running time, GA-BPNN model and PSO-BPNN model followed, while the proposed FA-BPNN model has the shortest running time. Therefore, the proposed FA-BPNN model in this paper is feasible and effective. Then, the proposed FA-BPNN model is used to improve the data fusion in WSN technologies. Finally, the improved WSN technologies are used to collect the actual traffic flow, but the cost is always very high. Therefore, the proposed FA-BPNN model is used to predict the actual traffic flow. Compared with the other three kinds of BPNN models, the proposed FA-BPNN model has the best effect and the shortest running time in the traffic flow prediction.