Study On Statistic Analysis And Prediction Model Of Road Traffic Accident

The rapid development of transportation accelerates the prosperity of economy, however, at the same time, results in the increase of traffic accident, and imperils the people's life and possession. It is urgent to find a method which can reduce the loss, prevent traffic accident and improve the status of traffic safety. To realize this target, mere amelioration of the basis establishment of road is not enough. Scientific supervision should also be required, thus Emergency and Security System is proposed. It integrates the advanced technology of information, communication, artificial intelligence and computer network, etc. into the whole traffic system, so it has a great significance in effective reduction of traffic jam and the accompanying loss, timely salvation of the wounded and amelioration of the traffic condition. This dissertation is based on one project of Science and Technology Committee in Chongqing named Emergency Management and Safety System of Urban Traffic. Considering the unique characteristics of Chongqing's geography and its real-time traffic conditions, the aim of the project is to design an Emergency Management and Safety System which can best apply to the true condition in Chongqing. The system can provide the prediction of traffic flow, road conditions and traffic accidents to the traffic supervisors, which will put forward intelligent emergency solution strategy and coordinate the relative sections, thereby advance the efficiency of accident management. The research on the statistic analysis and the prediction model of road accident is an important part of the system. Through the statistic analysis of various factors which may lead to the accident, it reveals the intrinsical rule and causes of accidents, based on which we can predict the future trend of accidents in order to provide reliable information for effective accident prevention. Firstly, the dissertation introduces the current situation both at home and abroad in the area of traffic safety, and analyses the causes of accident from several aspects, such as driver, vehicle, and surroundings based on the investigation of the traffic accidents in Chongqing. It expatiates on the correlation between the accident and its causes, and lays the foundations for the prediction model. Secondly, some prevalent theories for accident prediction are presented, such as regression analysis, time serial prediction, gray prediction and neural network prediction, etc. And then the theory of Bayesian Network is introduced as the modeling algorithm adopted in this research: on the basis of analysis of accident causes, the historical accident data are used to construct the network model, which produces the occurrence probability of accidents by calculating the conditional probability among variables, and accordingly achieves the target of accident prediction. Finally through the simulation of historical data, it proves that the proposed algorithm is efficient and feasible. The dissertation also analyzes the practical situations to which the model can be applied.