A modified recurrent least squares algorithm for online closed-loop identification for processes with time delay

Focuses on an algorithm of linear digital dynamic models online identification based on recurrent parameter estimation for processes with time delay. This algorithm is efficient for adaptive control systems design. The delay of the linear part of a digital model in the S-domain is estimated as well as parameters of a numerator and denominator of a transfer function (TF) in the Z-domain. This allows us to correct the delay of a model in the Z-domain and accordingly to make a correction of the recurrent least squares algorithm. The experimental results of prop-fan engine identification are submitted. The algorithm can also be useful for checking of process model structure correctness and choosing of an optimum interval of time discretisation in digital control system design.