CHAOS NEURAL NETWORK THEORY BASED MODEL FOR QUANTITATIVE PRED ICTION OF URBAN GROUND SUBS IDENCE

urban ground subsidence is of nonlinear dynamic character and its quantity in time series is analyzed in this paper. Then it is assumed that there is chaos in the urban ground subsidence in time series. Based on this as2 sump tion and using the chaos neural network theory, a p rediction model of urban ground subsidence quantity was built with phase space reconstruction. The ground subsidence quantity in Gao - qiao analyzed and p redicted with thismode1. The observed data are compared with the p redicted data. The largest absolute p rediction error is 1. 7 and the average forecast error is 0. 0833. The results indicate that the chaos neural network theory is reasonable and accurate to p redict the urban ground subsidence.