Modelling and hierarchical learning control of a dryer

This paper deals with modelling and hierarchical learning control of an industrial phosphate dryer. The model is derived from heat and mass balances. It consists of hyperbolic nonlinear partial differential equations. The control and the output variables appear at the boundary conditions. The method of characteristics that is based on two independent variables is used for numerical simulation purposes. The control objectives are to minimize fuel consumption and to keep the moisture content of dried phosphate close to a desired value, despite external perturbations acting on the drying process. The fuel flow and the dried product moisture content have been selected as control variables. A hierarchical learning system, operating in a random environment that corresponds to the dryer, is used for control purposes. During its operation the learning system collects the pertinent information about the variables that describe the process to be controlled and generates a control action. The obtained results show the good performance characteristics of the considered controller.