Title :
Identification of extended Hammerstein systems for modeling sticky control valves under general types of input signals
Author :
Fang Lei ; Wang Jiandong ; Zhang Qinghua
Author_Institution :
Coll. of Eng., Peking Univ., Beijing, China
Abstract :
This paper is motivated by modeling the control valve stiction phenomena in feedback control loops. An extended Hammerstein system with discretized Preisach model as the input nonlinearity is proposed to capture hysteretic dynamic behaviors of cascading systems consisting of control valves followed by linear dynamic subsystems. An iterative method is presented to identify the extended Hammerstein systems. An informative input needs to have multiple local extreme values in order to ensure the discretized Preisach model identifiable. However, if only a single hysteresis loop of the discretized Preisach model is in active, then an oscillatory input with only two extreme values is shown to be persistently excited for identification. Hence, the proposed identification method for extended Hammerstein systems is applicable to general types of inputs as well as oscillatory ones. Two industrial examples are provided to demonstrate the effectiveness of the proposed identification method for extended Hammerstein systems.
Keywords :
cascade systems; control nonlinearities; feedback; hysteresis; identification; iterative methods; linear systems; stiction; valves; cascading systems; control valve stiction phenomena; discretized Preisach model; extended Hammerstein systems; feedback control loops; hysteretic dynamic behaviors; input nonlinearity; input signals; iterative method; linear dynamic subsystems; multiple local extreme values; oscillatory ones; single hysteresis loop; sticky control valves; Estimation; Feedback control; Hysteresis; Iterative methods; Relays; Valves; Vectors; Control valve stiction; Hammerstein system; Nonlinear system identification; Persistent excitation; Preisach model;
Conference_Titel :
Control Conference (CCC), 2013 32nd Chinese
Conference_Location :
Xi´an