Main Article Content

Karthick SHANMUGAM
Kanthalakshmi SRINIVASAN

Abstract

Optimal control is mainly concerned in operating the system with minimum cost. The most promising optimal control strategy available in literature is linear quadratic regulator (LQR). In LQR, it is important to select the state (Q) and control (R) weighting matrices to get optimal results. With no standard guideline for selection of these weighting matrices, the generally adopted trial and error method makes the job of a control engineer more tedious and tiresome. To address this issue, a hybrid particle swarm optimization algorithm (HPSO) to obtain optimal weighting matrices is proposed in this paper. Moreover, the premature convergence of the particles leading to suboptimal results is eliminated by introducing a local convergence monitor, which transforms the entire population at the occurrence of local convergence to a new search space. The proposed HPSO tuned LQR control strategy is applied to cart position tracking and pendulum angle regulatory control of a single inverted pendulum, which is a highly nonlinear unstable system. Experimental results reveal that compared to PSO tuned LQR, HPSO tuned LQR has improved tracking response with smooth error convergence.

Article Details