Abstract
In Japan, there are many mountain districts, and land use around the steep inclination area proceeds remarkably, in addition, because of much rainfall, many land slide disasters occur every year. The limit rainfall curve in which variables are unit time rainfall and accumulation rainfall is proposed as an index of the land slide risk in rain. But, the limit rainfall curve is decided uniquely regardless of the ground condition.
In this study, in order to establish the prediction method of slope stability in rain, 2 dimensional FEM analysis in which input variables are the total and unit time rainfall and the soil permeability was done. The method which decided the limit rainfall curve that a slide safety factor became 1 was proposed by neural network that regress the input variable and the safety factor in non-liner shape. And we quantitatively evaluated the influence to the safety factor by the inconstant rain in which unit time rainfall changed.
