A type of Time-Varying First-Order Plus Dead-Time (TV-FOPDT) model is extended from SISO format into a MISO version by explicitly taking the disturbance input into consideration. Correspondingly, a set of on-line parameter identification algorithms oriented to MISO TV-FOPDT model are proposed based on the Mixed-Integer-Nonlinear Programming, Least-Mean-Square and sliding window techniques. The proposed approaches can simultaneously estimate the time-dependent system parameters, as well as the unknown disturbance input if it is the case, in an on-line manner. The proposed concepts and algorithms are firstly illustrated through a numerical example, and then applied to investigate transient superheat dynamic modeling in a supermarket refrigeration system.
Ifac Proceedings Volumes, 2013, p. 563-568
Main Research Area:
: 13th IFAC Symposium on Large Scale Complex Systems: Theory and Applications, 2013