Controller tuning using evolutionary multi-objective optimisation: Current trends and applications

• We show current applications of evolutionary multiobjective optimisation (EMO).
• We focus on PID, predictive, fuzzy and state space feedback controller tuning.
• EMO has to be employed in a holistic multiobjective optimisation design procedure.
• We identify possible trends and research topics of EMO for controller tuning.

Control engineering problems are generally multi-objective problems; meaning that there are several specifications and requirements that must be fulfilled. A traditional approach for calculating a solution with the desired trade-off is to define an optimisation statement. Multi-objective optimisation techniques deal with this problem from a particular perspective and search for a set of potentially preferable solutions; the designer may then analyse the trade-offs among them, and select the best solution according to his/her preferences. In this paper, this design procedure based on evolutionary multiobjective optimisation (EMO) is presented and significant applications on controller tuning are discussed. Throughout this paper it is noticeable that EMO research has been developing towards different optimisation statements, but these statements are not commonly used in controller tuning. Gaps between EMO research and EMO applications on controller tuning are therefore detected and suggested as potential trends for research.