Master-Slave Stochastic Optimization for Model-Free Controller Tuning


ATEŞ A., ALAGÖZ B. B., YEROĞLU C.

IRANIAN JOURNAL OF SCIENCE AND TECHNOLOGY-TRANSACTIONS OF ELECTRICAL ENGINEERING, cilt.41, sa.2, ss.153-163, 2017 (SCI-Expanded) identifier identifier

Özet

This paper presents the model-free tuning of Fractional Order Proportional-Integral-Derivative (FOPID) controllers for twin rotor multi input multi output system (TRMS) using master-slave Stochastic Multi-parameters Divergence Optimization (SMDO) method. Second Order System (SOS) is used as a reference model for guidance of tuning process. Master and slave SMDOs perform optimization of design coefficients for the master system SOS and the slave system FOPID controller in parallel. In the first stage of optimization, called locking, step responses of FOPID and SOS approximate to each other. In the second stage, called drifting, SOS guides the tuning of FOPID control towards a desired step response of master system SOS. We observed that the proposed method provides two major advantages of master-slave SMDO method: First, the optimization of reference model (SOS) and FOPID controller is performed together and this allows better fitting of the reference model and real control system without a prior model assumption. Second, this optimization strategy can be effective for online fine-tuning of real FOPID control system to match a suitable SOS dynamics.