An application of model-based fuzzy control to an industrial grain dryer

Although many fuzzy controller applications have been reported, these have bean exclusively rule-based designs. This paper describes an application of a model-predictive controller, based around a fuzzy relational model, to a grain dryer in a Whisky distillery. The dryer was a particularly difficult system to control since it had a significant dead-time, poorly understood internal mechanisms and large non-linear disturbances. The paper describes the process and the methods used to gather data for identification. The data analysis techniques and methods used to construct the fuzzy relational model are also described. Finally, the results of on-line controller trials are presented and these show that good control performance is achieved, especially when compared with the previous manually controlled system.