Analysis On Driver's Braking Behavior Characteristic In The Simulated Rear-end Collision Avoidance Scene

Rear-end collision is one of the most important accident forms. A great many rear-end collision take place all over the world every year, going with serious casualties and enormous wealth loss. Therefore, study on driver's behavior characteristics in rear-end collision avoidance is important and meaningful for the road traffic safety study field.According to the road traffic developing requirement and the important role driver plays in the traffic system, the purpose of the study is to reveal the interrelation and rule between driver and road traffic safety in a given time domain and space domain, in order to programming and correspond the configuration, capability and performance of every component and make it optimized, study was carried out by doing as follows:Road traffic accidents have been one of the social effects of pollution all over the world currently. Enormous social losing is made in the vehicle accident in the past hundreds of years. The problem of traffic safety has already attracted more and more attention of the social. Because of the important role that driver plays in the traffic accidents, academicians have devoted much into the field of driving behavior characteristics study.Driving behavior characteristic is an important part of the related study on traffic safety, various of research have been carried out by the expert in this field all over the world, and driver's brake characteristic is during end-collision avoidance is one of the most important branches. End-collision is a familiar form in road traffic accident. End-collision happens with high frequency every year in every country and has made enormous social losing and economy losing. Therefore, it has an emphasis meaning for the road traffic safety to carry out the study on driving behavior characteristic during the end-collision avoidance.The moving state and its change trend of the vehicle during the end-collision avoidance change easily. There are some certain differences of driver's judgment and decision between the ending-collision avoidance and general driving process. Therefore, it's difficult to estimate and modeling particularity, impersonality, accurately on driver's brake behavior characteristics during ending-collision avoidance by the traditional study methods.On base of the current internal and external research results, analysis of traditional study theory and method on driving behavior characteristic was carried out. Neural Network model of driver's brake behavior characteristics during ending-collision avoidance was set up. The driving behavior test under the scene of the simulated traffic environment was carried out on the fixed driving simulator improved by the State Key Laboratory of Automobile Dynamic Simulation of Jilin University. Neural Network GUI was set by Matlab to select the appropriate network structure and learning arithmetic, and finally the driver's brake behavior characteristic model was built up for the simulated ending-collision avoidance scene. PC-Crash was cited to verify the practice example. It can be concluded that the Neural Network model can simulate driver's brake behavior characteristics during ending-collision avoidance exactly and can be used as assistant liability affirmation way of the traffic accident.According to the road traffic developing requirement and the important role driver plays in the traffic system, the purpose of the study is to reveal the interrelation and rule between driver and road traffic safety in a given time domain and space domain, in order to programming and correspond the configuration, capability and performance of every component and make it optimized, study was carried out by doing as follows:1. The important role driver plays in the traffic system and the necessary of the driver's behavior study is analyzed and do research on driver's judging and decision-making process. According to the research experience at home and abroad, and the collision avoidance regular in navigate and aviation was cited in the study idea, quantified driver's characteristic is proposed as behavior assistant evaluation standards.2. Based on previous experience, test method in the regular driving behavior characteristic study, and driving simulator as the test platform, simulated traffic situation is established, the repeatability of the test and the accuracy of the data collection is enhanced.3. The GUI module of the Neural Network is set up to optimize the network configuration and the learning algorithm, and form the final driving behavior NN model. By NN training, relationship of between the relative motion state of the vehicles and the judgment, decision-making of driver was generalized. The NN after training was used to simulate driver's practically operation.4. Driving behavior varying process can be more directly shown by vehicle movement state calculation module of the PC-Crash. Contrast the sample and the simulated result, and ideal verification effect was revealed.5. The method and model in the study was used in a traffic accident analysis. Driver's behavior in the accident was analyzed. The driver's operation process and the liability were judged. The results indicate that, the research method offered in the study can be used in the assistant cognizance of the accident liability.The research in the paper will prompt the development of optimization study on performance of driver in the traffic system. As the complicacy of the traffic accident and the variety of the vehicle, the driving behavior characteristic neural network model found in the paper need to be perfected. For example, abound the training sample data-base and the vehicle model data-base, and make up the interface plat-form during the different modules.