Biased-Relay Feedback Identification for Time Delay Systems

The paper describes a new technique for estimation up to three points on a process frequency response from a single biased-relay feedback test without using the FFT algorithm. The new technique called "shifting method" extends possibilities of the relay feedback identification of dynamic systems describable by linear models. The introduced identification method is applicable for systems with dead time if there are stable limit cycles in biased-relay feedback tests. This approach enables to fit up to five parameters of a process transfer function from data observed by one relay experiment. The method may be used for parameter estimation of linear mathematical models with various structures and therefore it can be used for parameter fitting both isochronic and anisochronic linear models. This is demonstrated on proportional and integrating time delay processes described by isochronic or anisochronic linear models. The paper also introduces explicit formulas for the parameter estimation of the process model of the second order plus time delay. The applicability of the proposed method is demonstrated on simulated examples. Matlab/Simulink programming environment was used for all simulations. (C) 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.