A HYBRID GA FOR NOX EMISSION MODELLING IN POWER GENERATION PLANTS

Abstract This paper reviews a grey-box (GB) modelling method that uses genetic algorithms (GA). The GA-GB modelling framework is used for finding NOx emission from coal-fired power generation boilers using operator controlled variables. The main contribution of this paper is the inclusion of a distributed elitist scheme within the GA. This enables a local gradient-based optimisation routine to be incorporated within the GA. The new hybrid GA-GB modelling procedure is shown to be able to produce NOx models having similar performance to those of the original method but requires less computational effort.