Design of adaptive laws for discrete-time systems based on Particle Swarm Optimization

A wide variety of linear/nonlinear adaptive systems in continuous/discrete time can be represented by error models, thus facilitating their analysis. The solution obtained for a given error model can be applied to different systems represented by the error model. This paper presents a methodology for adjusting the parameters of a discrete-time adaptive system represented by a Type 1 Error Model, which is based on Particle Swarm Optimization (PSO) and that allows using it in on-line applications. The performance of the proposed methodology is compared with the traditional gradient and least squares methods through simulations.