Prediction of the time dependent in-situ pressure of soft rock using multiple regression approach, artificial neural network, and adaptive network-fuzzy inference system

摘要

Soft ground is a non-linear material with time-dependent characteristics. It causes major prob-lems of supporting both during construction and during the operational life of an underground excavation due to swelling and squeezing phenomenon. Therefore, it is worth monitoring the swelling and squeezing behavior of soft grounds in underground excavations. Compensation method is one of the most famous methods for determining the in-situ pressure of such ground on support systems. The study presented herein aims to predict the variant stress of concrete lining due to time dependent pressure of soft rock based on the closure of pine distances before and after making the slots in compensation method. In order to establish predictive models, statistical and soft computing techniques such as multiple regression approach (MRA), artificial neural network (ANN) and adaptive network fuzzy inference system (ANFIS) were used, and prediction performances were then analyzed.