Constrained Model Predictive Control of an Industrial Drying Process for Viscose Staple Fibres

The paper presents the implementation of a model based predictive control algorithm of an industrial drying process for Viscose staple fibres. A short review of the earlier published modelling procedure presents the linearised multi-input multi-output (MIMO) state space formulation of the design model. The design of the predictive control scheme incorporates constraints on the manipulated variables, set-points for the manipulated variables, and active disturbance rejection for the main disturbance variables of the drying process. Some simulation results demonstrate the beneficial effects of this structure, and the choice of proper weightings is also briefly discussed. Implementation issues and various measured results are presented. The problem of redundant but biased measurements of fibre humidity is addressed and the superior performance of the new concept compared to the existing PID-control is demonstrated. Both energy consumption and variation of humidity in the final product are significantly reduced by the successful implementation.