G.A. Optimisation of PID Controllers — Optimal Fitness Functions

Genetic Algorithms (GAs) have been in many cases successfully applied to a wide variety of optimisation problems. The work described here focuses on the application of genetic algorithms to the optimisation of linear & non-linear PID controllers. The techniques used here are based around the formulation of a suitable objective function, which, as part of the GA evaluates the fitness of a given PID parameter set. The proposed objective function is based directly on performance criteria specified in terms of the rise time, settling time and peak overshoot of a physical system. The functions are evaluated via a set of trials on a range of systems with varying dynamics and the success rate determined by comparison with sets of target step response characteristics. The results show that the objective function is more robust than ISE based methods in the optimisation of multiple step response objectives, having a low deviation from the target across the range of parameters (rise time, settling time & peak overshoot).