Study of Bus Incident Prediction Based on Dynamic Fuzzy-Neural Network

Bus incident are random, while its prediction is a nonlinear time-varying system, and is difficult to use mathematical models to achieve accurate modeling .The paper presents a real-time forecasting method of bus incident based on a dynamic fuzzy-neural network .First, the paper defines what are called bus incident, and then analyzes the factors that will lead to bus incident. Introduce using cluster analysis and stepwise regression methods to predict bus station passenger traffic, and using Fuzzy C means cluster analysis method to identify road traffic state. Based on the above prediction results include bus station passenger traffic and road traffic state. the paper constructs the network of dynamic fuzzy-neural to predict the bus incident, and simulates the network based on Matlab simulation software. Simulation results show that the speed of network training is fast, the prediction result is high accuracy and with fine real-time. The simulation results also show that dynamic fuzzy-neural network is a good method to predict bus incident.