Selection of reduction parameters for complex plant MIMO LTI models using the evolutionary algorithm

This paper discusses the optimal selection of reduction parameters for complex plant models for the Frequency Weighted and the Frequency Domain method together with the Evolutionary Algorithm. The algorithms presented here enable precise reduction of linear time invariant (LTI), multi-input and multi-output (MIMO) models and are particularly useful for models in which the individual channels differ significantly from each other and are characterized by a specific scope of adequacy for the frequency domain. A two-level reduction algorithm was introduced in order to shorten the computation time of the reduction of very high-order models. The algorithms are then applied to the reduction of the models of the BP-1150 steam boiler subsystems.