An Improved MRAC Scheme for Non-linear Design of Adaption Gain γ using Heuristic Algorithms

Abstract

This paper presents an approach to Optimal Model Reference Adaptive Control (OMRAC) and comparative performance evaluation are analyzed for a conical tank process which exhibits behavior of a class of nonlinear dynamical system. The conventional Model Reference Adaptive Control (MRAC) method based on the MIT rule reported in the works (P. Swarnkar et al., 2010), has its difficulty in choosing the reference model and adaption gain γ. Inspired by the above work, the selection of reference models in MRAC scheme are based on the multiple models (MM) depending on the operating regime of the process. In this proposed work, the reference model considered is a second order system having fixed roots in denominator polynomial representing the most dominant poles in the process determined using Dominant Pole Algorithm (DPA). The numerator polynomial of the reference model is designated as static sensitivity from each linearized region of the process. The optimal adaption gain γ of MRAC is obtained by Bacterial Foraging based Particle Swarm Optimization (BFPSO) algorithm and the system performance is compared with traditional algorithms such as Bacterial Foraging Optimization (BFO) and Particle Swarm Optimization (PSO). The simulated result gives consistently better setpoint tracking mechanism and error minimization.