Finite-horizon optimal LQ control design using artificial bee colony programming

Abstract

Designing optimal control laws in closed analytical form is still showing challenging computational issues. This may even hold for moderately complex problems like the linear quadratic (LQ) control problem with finite horizon. This paper introduces a novel approach to LQ controller design based on artificial bee colony programming (ABCP). Solution to the matrix Riccati differential equation (MRDE) derived for the optimal control problem subject to linear dynamical system model is determined by means of ABCP method. Aiming at this, preparatory steps have been set and analyzed as essential part of the ABCP-based MRDE solver. The effectiveness of the proposed solver is investigated on a typical LQ control problem and compared to existing methods in literature. Evidence of superiority is shown through numerical evaluation of the method convergence and solution accuracy