This paper reports on our present achievement toward the intelligent control of a boiler-turbine power-plant based on switching control scheme, recently revived by some active reports. To overcome strong nonlinearity emerging in load following operations of boiler-turbine power plants, which is not efficiently compensated by the conventional PI-based gain scheduling control, a neural-based nonlinear feed-forward switching control scheme is employed. Owing to its 2-degree freedom type installment in the control system and proper switching of nonlinear feed-forward control by monitoring contribution of inverse dynamics error to control error, effective suppression of nonlinearity is achieved.
a subsidiary of the Penn Engineering GRASP Lab