DYNAMIC MATRIX CONTROL WITH EXPERT KNOWLEDGE-BASED MULTI-LAYERED HEURISTIC SELF-TUNING

A new dynamic matrix control (DMC) algorithm with expert knowledge-based multi-layered heuristic self-tuning is presented in this paper. In the algorithm, the correction strategy is pivoted on real-time outputerror and internal model error in the higher level, while in the lower level model identification self-tuning or heuristic self-tuning DMC is employed according to the extent of model mismatch. The adjustment experience of experts and conventional identification self-tuning method are merged into a whole structure, therefore the dynamic performance of closed-loop system with model mismatch is improved.