Fuzzy neural networks in nonlinear system identification

There is much interest in closed-loop system identification recently. In this paper, based on historical input output data, we construct initial system model using fuzzy method in order to solve the identifiability of closed-loop systems. The initial model is then modified based on present time data using OLS learning algorithm in order to enhance the precision. The new identification method is used in real data from an ammonia process with satisfactory results.