Water-Destroyed Disaster Forecast of Expansive Soil Roadbed under Extreme Climate Conditions

Water-destroyed of Roadbed on expansive area was gravely affected as extreme climate when the globe is warming. The Pearson Type III distributing model was established which can imitate the daily rainfall extremum. The Feng Lihua's drought model was introduced to evaluate the drought scale. The fine-rain translation matrix was established based on Markov Chain. And the Rough Neural Network was adopted to forecast the disaster of expansive soil roadbed water-destroyed under extreme climate conditions. Due to different outputs for same input in a trained Neural Net, the Monte-Carlo method was presented to simulate the probability of swelling and shrinkage grades, rainfall, drought continuance and roadbed gradient. The probability was inputted to the trained Neural Net. The statistical result of the output was the water-destroyed probability of the expansive soils roadbed. The calculating anlysis indicates this arithmetic is an expandable one which can calculate water-destroyed probability of the expansive soils roadbed when its geography distributing coupling with extreme climate and this arithmetic can be also used to forecast other disaster under extreme climate conditions.