MIMO CCS model identification based on GA for closed-loop field data

Abstract

The MIMO model identification methods based on experiment closed-loop field data for power plant boiler-turbine coordinated control system (CCS) is developed in this paper. The CCS fast and slow dynamic processes, which correspond to large and small time constants with time delay, are modelled by combining the genetic algorithm (GA) with step response analysis. The step response is obtained from the closed-loop system response tested by normal operation signals. The step modelling formula is derived, and the identified parameters are used to initialise and to form the ranges for GA chromosomes. This multivariable system identification is suitable for control system parameter tuning of modern power plant equipped with distributed control system (DCS). Some simulations and considerations for the stable and margin stable system application of the technology are also reported.