Neural control for NOx emissions in a sludge combustion process

Abstract

In this paper, a discrete-time neural control scheme to regulate carbon monoxide (CO) and nitrogen oxides (NOx) emissions for a fluidized bed sludge incinerator is proposed. Carbon monoxide emissions are reduce by oxygen regulation in the incinerator; nevertheless nitrogen oxides emissions are difficult to control because the sludge composition varies continuously. This scheme ensures carbon monoxide and nitrogen oxides regulation without decreasing combustion efficiency. In order to obtain the sludge combustion model, it is proposed to use a recurrent high order neural network (RHONN), which is trained with an extended Kalman filter (EKF) algorithm. The proposed neural controller performance is illustrated via simulations.