Optimal Design of LQR Weighting Matrices based on Intelligent Optimization Methods

In this paper, considering some important indices such as closed-loop pole locations, speed of response, and maximum level of control effort, and combining them into an objective function, an optimization problem is defined to find the optimal weighting matrices in LQR controller. To solve this optimization problem four intelligent optimization methods are utilized: Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE), and Imperialist Competitive Algorithm (ICA). The proposed method is applied to a nonlinear flexible robot manipulator model, and obtained results from the algorithms are compared.