Plant identification by relay method

Plant identification is very important in control design. The relay method belongs among special kinds of experimental frequency response identification. It was originally used for plant identification by Rotach (Rotač, 1964) and lately also for controller autotuning, see reference (Åstrőm & Wittenmark, 1989; Åstrőm, 1993; Åstrőm & Hägglund, 1995; Bi et al., 1997; Hang et al., 1993; Lu et al., 1998; Shen et al., 1999; Tan et al., 1996; Wang et al., 1999). Autotuning of controllers for MIMO plants is described in references (Halevi et al., 1997; Loh & Vasnani, 1994; Semino & Scali, 1998). At present more publications exist which are devoted to identification or controller autotuning on the basis of the relay experiment, e.g. (Besançon-Voda & Roux-Buisson, 1997; Hang et al., 2002; Huang et al. 2005; Johnson & Moradi, 2005; Kaya & Atherton, 2001; Lee et al., 1995; Leva & Donida, 2009; Leva & Maggio, 2009; Liu & Gao, 2009; Majhi, 2005; Majhi, 2007; Panda & Yu, 2003; Panda & Yu, 2005; Prokop et al., 2010; Sung et al., 2009; Tan et al., 2001; Tan et al., 2002; Vítečková & Víteček, 2005; Vivek & Chidambaram, 2005). The aim of this paper is to summarily describe and show the basic modifications of the relay methods from the viewpoint of experimental plant identification and to bring out the computational formulas for simple plants. Two-position symmetric relays without and with hysteresis and with the integrator in front of the relay and behind of the relay are considered.