基于BPNN的高速公路风雨天气限速系统的设计

摘要

为了实现高速公路能够在风雨条件下智能限速，作者采用反传播神经网络（BPNN）技术，设计了一个高速公路风雨天气限速系统。由经验丰富的驾驶员根据交通环境，参考实时天气指标，凭经验判断“安全限速”的值，并作为专家样本，供 BPNN 训练。通过分析“拟合误差”的方法，排除了少量异常的专家样本，从而提高了BPNN的预测精度。实践表明，采用BPNN对高速公路按天气指标智能限速，无需推理数学模型，操作效率高，成本低，便于推广。%This study adopted the BP neural network technology to design a highway speed control system for stormy weather. First, the experienced drivers determined the values of “safe speed limit” which was used as the expert samples according to their experience and based on the on-site traffic environment and the real-time weather indicators. The expert samples were used to train the neural network. Then, by applying the“fitting error”method, a small number of abnormal expert samples were excluded to improve the prediction accuracy of the neural network. The practical results showed that by adopting the neural network to intelligently control the highway speed according to the weather indicators, it was unnecessary to derive the mathematical models any more. The operating efficiency is much higher while the cost is much lower, and hence, this system can be easily promoted.