Comparison of four methods of on-line identification and controller tuning

The work compares on-line methods of identification and controller tuning when the extreme points of the closed-loop response are either accurately known or subject to error. It is shown that Chen´s (1987) and Lee´s (1989) methods are relatively robust, with Chen´s method giving slightly better estimates of process ultimate data. The method of Jutan and Rodriguez (1984) gives acceptable results and Yuwana and Seborg´s method (1982) is the least acceptable. The work underscores the need to use high feedback gain during identification when experimental data are poor.