Multi-objective genetic optimisation of GPC and SOFLC tuning parameters using a fuzzy-based ranking method

A multi-objective genetic algorithm is developed for optimising the tuning parameters relating to the generalised predictive control (GPC) and performance index table of the self-organising fuzzy logic (SOFLC) algorithms, using a multi-objective ranking method based on fuzzy logic theory. A comparative study with more traditional pareto, average and minimum distance ranking methods shows that the proposed method is superior. The study shows that the approach leads to a more effective set of tuning parameters, especially those relating to the important observer polynomial for GPC and to a good reference trajectory for SOFLC. Up to two objective functions were used in the study, although the method can be extended to more objectives. A nonlinear muscle-relaxant anaesthesia model is used as a case study to demonstrate the robustness of the method.