点击次数：

论文类型：期刊论文

发表时间：2014-03-01

发表刊物：JOURNAL OF CENTRAL SOUTH UNIVERSITY

收录刊物：SCIE、EI

卷号：21

期号：3

页面范围：1206-1216

ISSN号：2095-2899

关键字：intelligent modeling; neural network; pressure balance system; excavation chamber; analytically modeling approach

摘要：In order to deal with modeling problem of a pressure balance system with time-delay, nonlinear, time-varying and uncertain characteristics, an intelligent modeling procedure is proposed, which is based on artificial neural network (ANN) and input-output data of the system during shield tunneling and can overcome the precision problem in mechanistic modeling (MM) approach. The computational results show that the training algorithm with Gauss-Newton optimization has fast convergent speed. The experimental investigation indicates that, compared with mechanistic modeling approach, intelligent modeling procedure can obviously increase the precision in both soil pressure fitting and forecasting period. The effectiveness and accuracy of proposed intelligent modeling procedure are verified in laboratory tests.