Parameter identification for a class of multivariable non-linear processes

Abstract A new signal-processing technique is presented and applied to the problem of determining the parameters of processes whoso dynamic behaviour is characterized by a set of ordinary first-order non-linear differential equations. The process signals and certain products of the process signals are used to excite Poisson filter chains. The Poisson filter chains provide exponentially smoothed moments of the signals at the filter chain inputs. Simultaneous sample values of these Poisson filtered signals are then used to form linear algebraic equations in the unknown process parameters. The method of instrumental variables is utilized to combat inaccuracies in the measured process signals. Finally, the method is illustrated by an example and the results of simulation studies are presented to demonstrate the feasibility of the method.