Research On The Active Control Strategy For Driving Safety Based On Information Service

In recent years, there is an increasing tendency of highway traffic accident, especially rear-end collisions frequently occur. Hence there is a growing concern over the driving Safety. Passive safety and active safety technologies are adopted to improve the traffic safety. Passive safety technologies can not effectively avoid accidents and only can reduce the loss to some extent. While active safety technologies can help vehicles avoid traffic accidents from the source. Therefore, it has important theoretical meaning and realistic meaning to study the active safety.Based on previous research work for vehicle active collision avoidance system, this paper is dedicated to the study of rear-end collision risk evaluation and active collision avoidance control. The main research contents are as follow:The existing collision warning systems provide a warning to the driver when there is an imminent collision, but there is no plenty of time for the driver to brake before collision happens. Thus a new rear-end collision risk evaluation model is proposed in our paper. First, we systematically analyze the influence of humans, vehicles, roads and environment on rear-end collisions. Next before evaluating the rear-end collision probability, the Bayesian network regarding rear-end collisions should be constructed from generated training data. And the data is generated by the simulation based on car-following system. According to the current state, then the state in the next time interval will be predicted with Kalman filter. Based on the constructed Bayesian network and predicted state in the next time interval, the collision probability can be evaluated. To simplify active control, the collision probability is divided into two levels, i.e. high or low. Then simulation experiment proves collision risk evaluation model is effective.If the collision probability in the next time interval is very high, a control unit will be launched to autonomously handle the mechanic parts of vehicles to avoid accidents. Otherwise, if the risk level is lower, the following vehicle does not need to send the warning to its driver. For active collision avoidance system, a variety of controllers have been formed like PID and sliding mode control which rely on the mathematical model. However, the mathematical model of vehicle brake system is not clear. Therefore fuzzy logic controller(FLC) is adopted to control vehicles in this paper. Nevertheless the performance of the FLC relies on the number of fuzzy control rules. Hence Genetic algorithm(GA) is used to optimize rules. If GA is used in real time optimization of fuzzy rules, the time for reaching the optimized value relies on the number of rules. Consequently, offline tuning of fuzzy rules is adopted. In our paper, the SIMULINK toolbox of MATLAB has been used to design of fuzzy Controller with Genetic Algorithm. Then the simulation results show that the fuzzy Controller meet the requirements of the collision avoidance.