IDENTIFICATION OF HAMMERSTEIN MODELS USING MULTIVARIATE STATISTICAL TOOLS

The iterative Narendra-Gallman algorithm (NGA) for the identification of a nonlinear system representable by the Hammerstein structure is ex tended to perform simultaneous structure determination and parameter e stimation of multivariable chemical process systems. The parameters of the linear system obtained in state space form using canonical correl ations analysis and the coefficients of the polynomial type nonlinear elements are alternately adjusted, until convergence, to obtain the mo del. The theory is illustrated using data from an experimental heat ex changer and the simulation example of a realistically complex acid-bas e neutralization tank.