Self-optimizing control structure selection via differentiation

A new approach using differentiation to select controlled variables for self-optimizing control [8] is proposed and applied to the evaporation process described in [7]. Based on the gradient function derived in the reduced space [1], new selection criteria, which indicate the sensitivities of the gradient function to disturbances and to implementation errors, have been derived for different controlled variables. The particular case study shows that the new selection criterion is more effective and reliable than the minimum singular value criterion proposed in [8].