Locally weighted regression for desulphurisation intelligent decision system modeling

Abstract Locally weighted regression (LWR) is a memory-based learning method which performs regression around a point of interest, which is useful for learning the rule of complex phenomenon and system. This paper studies the possibility of using locally weighted regression for modelling an intelligent decision system for desulphurisation in metallurgical process and proposes a hybrid algorithm by combining LWR with Genetic Algorithm (GA). The proposed algorithm proves to be effective and practicable in its application.