Abstract
This paper proposes a tuning method of PID controllers for interference systems. PID control has various advantages on practical use with high robustness, needlessness of plant model, and has been a major method in industrial world. In addition, PID control is used in various forms such as multi-loop, decentralized controls and so on. However, there is disadvantage on taking a long time to tune the PID parameters because of trial and error approach to tune them. Especially in case of multi-PID controllers tuning in interference systems such as MIMO or multi-loop systems, the tuning time will be increasingly longer. Recently, several systematic PID tuning methods have been proposed, but their applicability to general systems has not been evaluated. In this paper, we present an efficient tuning method of the multi-PID controllers installed in an interference system using Simultaneous Perturbation Stochastic Approximation (SPSA) approach. The SPSA is based on a highly efficient gradient approximation requiring only two evaluation of loss function independent of the number of parameters being estimated. At first, in this paper, the outline of the SPSA algorithm is explained. Next, the problem setting and the tuning method of the PID controllers are described. Then the SPSA stochastic characteristics on tuning process of control system are presented during tune. Finally, the control performance is evaluated.
