Influence of hyper-parameters in algorithms based on Differential Evolution for the adjustment of PID-type controllers in SISO processes through mono and multi-objective optimisation

Abstract:

PID Controllers remain as the reliable front-line solution in feedback control loops. Even when their simplicity is one of the main reasons for this, the right tuning of their parameters is essential to guarantee their performance. As consequence, several tuning methods are available. Nowadays performing a tuning process via stochastic optimisation is an attractive solution for complex processes. Nevertheless, the solution obtained using such optimisation methods is very sensitive to the hyper-parameters used. In this paper, we propose to designers a set of hyper-parameters for diirerent algorithms based on Dilferential Evolution in SISO processes. Obtained results show several aspects to consider regarding the most promising values for several optimisation instances, facilitating knowledge transfer for new optimisation instances.

Año de publicación:

2023

Keywords:

• Evolutionary algorithms
• PID tuning
• Hyper-paramelers tuning
• Optimisation

Fuente:

scopus