基于神经网络的道路季节分类

Seasonal Roadway Classification Based on Neural Network

为了合理确定路面结构设计时的输入参数,引入自组织特征映射神经网络,结合Matlab软件对神经网络进行权值训练,将网络训练是否收敛来作为分类的依据,根据温度、交通量和降雨量等几个重要参数对道路进行季节分类,最后按照分类结果进行路面结构分析与材料设计.实践证明,该方法分类效果良好,能很好地解决路面设计参数的合理确定问题,从而大大延长路面的使用寿命,提高道路投资的经济效益.

In order to reasonably determine the input parameters in pavement structure design, a self-organized feature mapping neural network is introduced and its weight is trained with Matlab. Then, by taking the convergence of the training as the classification rule, a seasonal roadway classification is made according to such important parameters as temperature, traffic and rainfall. Moreover, pavement structure analysis and material design are performed according to the classification results. The proposed classification method is proved effective in determining reasonable design parameters of pavement. Thus, it greatly prolongs the service life of pavement and improves the economic benefit of road investment.