基于BP神经网络的车祸库预警技术

Accident Database Early Warning Based on BP Neural Network

针对如何有效预测车祸发生的可能性,从而达到车祸预警的目的,提出了一种新的车祸预警方法:通过建立车辆的车祸库,并结合BP神经网络技术达到车祸库预警目的。先构建合适的BP神经网络,再用车祸特征信息训练BP神经网络,训练好的BP神经网络就具有判断发生该类型车祸可能性的能力,最后把车辆行驶信息输入到已训练好的BP神经网络,就可以预测发生该类型车祸的可能性。用Matlab7.0.1进行了该方法的仿真实验,仿真结果表明该方法具有一定的可行性和有效性。

In order to achieve the purpose of accident early warning, this paper presents a new method： establish the vehicle accident databases, and combine it with BP neural network technology. First, construct a suitable BP neural network. Second, use the accident feature information to train the BP neural network, then the trained BP neural network can determine the possibility of this specific car accident. At last, send the vehicle information into the trained BP neural network, and it can predict the possibility of this specific car accident. In this paper, the author simulates this method by MatlabT.0.1. Simulation results show that the method is feasible and effective.