Highway Traffic Accident Detection Method with Fuzzy Neural Network

It is very important to detect highway accident through the computer control and traffic data detection. The detection method for highway traffic accident was developed with five layer fuzzy neural network (FNN in short) considering the self-organization map method and feedback error algorithm. The learning progress of FNN was analyzed, the original subordinate function was determined with self-organization model. The accident detection model was established with FNN theory. The network learning and training were accomplished with self-organization map method based on the traffic flow simulation data detected by single detection circle on the highway covering 3.6 kilometers. The output of the FNN shows that the decision can be determined at once when the accident happened. All the method, algorithm and theory were carried out with software to accomplish the traffic accident detection. The detection experimental results show that the detection method is efficient and robust.