Adaptive optimal control of steam temperatures for thermal power plants

An adaptive optimal control method using a process control computer was developed for boiler steam temperature control for thermal power plants. A process dynamics model, which describes the dynamic plant behavior, is used in the calculations for adaptive optimal control. The mode used is a multi-input and output ARMA (autoregressive moving-average) with parameters identified online by the Kalman filter method. The adaptive optimal control method was applied to a coal-fired thermal power plant simulator. Errors in predictions of the present model were one-third to one-half those of a model using parameters identified offline by the LWR (Levinson-Wiggins-Robinson) method. The fluctuation of the temperatures from the set points was reduced to half of that obtained by a conventional control method which uses PI (proportional+integral) controllers