The Study On Target State Recognition In Vehicle Rear-end Collision Avoidance Control System

The traffic accident maintains a high level, especially the rear-end collision accident accounts for a large proportion, which becomes an important factor of constraining the development of the expressway in China. Vehicle collision avoidance system is an active safety measure to radically solve traffic accident problem and reduce the loss. Multi-source information fusion technology can effectively use multi-sensor resources and could be a greater access to obtain the information of the environment and target to be detected. The vehicle rear-end collision avoidance control system based on the information fusion can get rid of the impact of traffic environment and driver's subjective factors, and has very positive implications of reducing traffic accidents, especially highway rear-end collision accidents. The further research for the target state recognition in the vehicle rear-end collision avoidance control system is carried out in this dissertation, and its main work as follows:(1) On the basis of the further analysis for the function model, structure model and key problems of the multi-source information fusion, the problems about the target state information fusion and state evaluation are deeply discussed.(2) The system structure of vehicle rear-end collision avoidance control system based on information fusion is given, and its principles and system structure are analyzed. Also the deep research and discussion for the key technologies such as vehicle driving environment information detection, sensor selection and target state recognition etc are carried out.(3) The radar-infrared sensor data fusion algorithm and its realization are given on the basis of the given information fusion algorithm based on the optimal data compress. The simulation results indicate that measurement results and target state recognition precision can be remarkably improved.(4) The application of neural network in the target state information fusion is researched, and the radar-infrared fusion algorithm and structure based on the BP and RBF network are built. The simulation results show that the target state recognition precision can be greatly improved by using the nonlinear function approximation ability of the neural network.