A least-squares like gradient method for discrete process identification

A new deterministic ‘ least-squares-like gradient’ method is presented for the identification of discrete processes. The method is gradient-based and physically similar to the recursive least-squares method. The novel gradient method is based on a stability concept (Liapunov's second method) yielding new views on the estimation procedure and. more degrees of freedom compared with least-sqviares methods. The method can be applied for linear and a class of non-linear (multivariate) processes with slowly time-varying unknown parameters.