Model-specific autotuning of classical regulators: a neural approach to structural identification

This paper presents a simple technique for the automatic tuning of classical regulators, based on a structural identification method. The core of the proposed approach is a neural pattern classifier, which selects an appropriate model structure for the system to be controlled, starting from a simplified representation of its unit step response. In addition, simple model-specific regulator synthesis rules are suggested which benefit from the available knowledge on the process model structure. The technique is compared with the classical IMC autotuning approach. A real application example, based on a laboratory distillation column, is also presented.