Improved MOGA-tuning and visualization for a hybrid control system

Authors:	Chipperfield Andrew, The University of Southampton, United Kingdom Stirrup Rodney, The University of Southampton, United Kingdom
Topic:	2.4 Optimal Control
Session:	Evolutionary Computation in Control Systems Engineering
Keywords:	Solar Thermal Power Plant, Multiobjective Genetic Algorithm (MOGA), Gain Schedule Control, Fuzzy Logic Control.

Abstract

A hybrid controller is developed for a solar-thermal power plant using a gain-scheduled controller with feedforward to control the more linear operating regimes and a fuzzy PI incremental controller for the highly nonlinear operating region of the plant. An enhanced method of MOGA-tuning is employed by first optimizing the number of input/output membership functions using neuro-fuzzy data clustering. Enhancements to the visualization properties of the MOGA’s graphical user interface are evaluated to improve the decision maker’s choice when deciding between non-dominated solutions or potential fuzzy controller inference systems.