中文摘要：

利用可辨识矩阵对影响路面使用性能评价的宽泛指标集进行分类、约简,得出对路面使用性能评价最有影响的数据指标,建立RBF神经网络模型,并把处理后的数据指标作为RBF神经网络的输入进行训练、仿真.通过实例,给出了该方法的具体实现过程.与没有采用指标约简的RBF神经网络进行结果对比,该方法在路面使用性能评价上更具有实用性、有效性和可靠性.

英文摘要：

In traditional highway pavement performance evaluation process,there are many disadvantages of method in using a single neural network,such as poor precision and low speed and so on.Therefore,a new way of highway pavement performance evaluation which uses rough set and RBF neural network was proposed.Firstly,it uses the discernible matrix to class and simplify the index which affects the pavement performance evaluation to get the most influential data index.Then it establishes the RBF neural network model to train and simulate by taking the data index of neural network as input.Finally,through the example,it gives the method to realize the process specifically.Compared with the RBF neural network which does not use the reductive index,this method has better practicability,validity and reliability in pavement performance evaluation.