路面状况指数的BP网络预测

BP Network algorithm of the prediction of pavement condition index

路面使用性能预测是路面养护策略决策的关键。在分析了路面使用性能预测模型发展中存在问题的基础上，利用神经网络建立路面状况指数ＰＣＩ的预测模型──ＢＰ网络模型。详细阐述了模型的变量选择和网络拓扑结构的建立，并以上海公路设施管理系统的数据为例对模型的应用和预测误差进行了分析说明。

The prediction of pavement performance is an important part of Pavement Management System(PMS). Although the traditional methods are widely used, they always can not reach the expected result for the complex axle-load and environment condition in the road. In this paper, a BP algorithm is used to create the prediction model of pavement performance. Moreover, variable and structure of BP choosing are given in detail. As an example for illustrating the calculated model and predicting its tolerance, the data of Shanghai Highways Infrastructure Management System were analyzed.