Model order formulation of a multivariable discrete system using a modified particle swarm optimization approach

This paper proposes an algorithm for model order formulation of an absolutely stable higher order linear time invariant multivariable discrete system using a new version of evolutionary computing technique namely, Modified Particle Swarm Optimization (MPSO). A simple adjunct polynomial method has been proposed for obtaining the initial seed values of the lower order multivariable system. In the modified PSO, the movement of a particle is governed by three behaviors namely, inertia, cognitive and social. The cognitive behavior helps the particle to remember its previously visited best position. This paper proposes to split the cognitive behavior into two sections. This modification is efficiently utilized to obtain a better lower order system that reflects the characteristics of the original higher order system by minimizing the integral squared error with the steady state constraints. The results obtained are compared with the earlier techniques utilized, to validate its ease of computation. The proposed algorithm is illustrated with a numerical example from the literature.

► Modified PSO based model order formulation technique was proposed in MIMO discrete system.
► Adjunct polynomial method gives the approximate second order model.
► Proposed lower order model closely follow the original model with minimum error
► Stability criteria is discussed by pole zero mapping.
► ISE of formulated order models are compared with general PSO and unified PSO techniques.