Road Traffic Accident Prediction Based On MATLAB Neural Network Model

The road traffic accident is one of worldwide environmental damages and it creates the serious damage of life and the property to the humanity every year. Around the world, our country is one of the countries which have the most road traffic accidents. Ten thousand vehicle mortality rates and 100,000 person of mortality rates have always been higher than the world average level. According to the statistics the vehicle inventory of our country has broken through 186,000,000 at the end of year 2009, and at the same time the automobile inventory has achieved 76,190,000. The vehicle inventory of our country occupies 8 percent of the world, while the traffic accident casualty actually stands the 16 percent of world, it has become one of serious social problems. The forecast of road traffic accident is one of the important contents of road traffic safe research, it has the vital practical significance on studying the road traffic accident's occurrence rule, analyzing the development of traffic accident's future trend under the existing road traffic conditions, regarding the road traffic security's plan and the decision-making.First taking the time as the main axle,this paper obtains the domestic and foreign road traffic survey from having a preliminary understanding of the road traffic accdient; Carrying on detailed analysis of the road traffic accident's reason from four aspects, the human (driver,passenger,pedestrian), the vehicles reason, the environmental factor, the path; Elaborating the common domestic and foreign road traffic accident forecast methods and analyzing their good and bad points by contrast; Moreover, it is difficult to using the function model to describing the road traffic accident and the reason causing it for their multi-factors, the complex misalignment relations characteristic.While the neural network has superiority on solving complex non-linear problem, has the distributional canned data, parallel learning capability, non-linear,better error-tolerance and so on,so we propose that we use the BP neural network model to carry on the road traffic accident's forecast. First analyzing many factors influencing road traffic accident's and carrying on normalized processing, then taking the most important factors as the input signal of the neural network through the interrelated analysis.The BP network model builds and trains the network by using the input and the output vector. Because it is difficult to realize the BP network model by computer programming language, while the MATLAB software enables this question to be solved for its strong function. Building the model through using the MATLAB neural network toolbox's function, this paper introduces the BP network model establishment process in detail. Finally it proves by using the actual data that this model may use in the road traffic accident forecasting.