Multivariable controller tuning using genetic algorithms for an induction motor

A methodology for tuning multivariable controllers in state space by using genetic algorithms is proposed. The multivariable controller is designed in discrete time by using an extended state space feedback with integral action. The feedback gain is computed by using a linear quadratic regulator. Optimal constraints are tuned by genetic algorithms according to a cost function. As a result a robust multivariable controller is obtained by using state space feedback. This tuning procedure avoids the knowledge of the exact parameters of the system, and has the advantage that reduces the tracking error and decrease the rise time based on a performance index. Proposed methodology is applied over an induction motor in discrete time. The performance of the proposed methodology is compared against fixed controllers calculated by linear quadratic methods.