膨胀土路基水毁灾害及其粗糙神经网络预测

通过对膨胀土地质灾害的总结，提出了膨胀土路基水毁灾害的分类分级方法，并根据常用的地质灾害预测办法，指出膨胀土路基水毁灾害的预测是一个典型的非线性问题，当考虑降雨、干旱等气候因素耦合影响时，宜进行危险性点评估。提出了采用粗糙神经网络对膨胀土路基水毁灾害进行评价和预测的方法，并运用粗糙集理论确定了该神经网络的结构。同时，以某膨胀土路基水毁大型模型试验的数据为例，进行了灾害预测，结果表明，将粗糙神经网络运用于膨胀土路基的水毁灾害预测是切实可行的。

Water-destroyed disaster of expansive soil roadbed and it's forecast based on rough neural network

The geologic disaster of expansive soil was summarized, and the method was presented to classify the water-destroyed disaster of expansive soil roadbed. The disaster forecast of expansive soil roadbed water-destroyed is a typical non-linear problem by analysis based on general geologic disaster forecast method. Fatalness evaluation must be point evaluation when the raion and drought were considered. The method was presented to forecast the waterdestroyed disaster of expansive soil roadbed by rough neural network. The structure of the neural network was confirmed by the rough set theory. The forecast was done with some data of one expansive soil large model test. The result shows it is feasible that bring the rough neural network to forecast the water-destroyed disaster of expansive soil roadbed.