THE NONPARAMETRIC PREDICTIVE CONTROL OF NON-MINIMUM PHASE PLANTS USING GENETIC ALGORITHM

Martina Leš - Rajko Svečko

   In this work, we have considered a non-parametric predictive control for systems with zeros outside the unit circle, known as phase non-minimal systems. The proposed algorithm is based on an impulse response and allows a control of all open-loop stable systems. We have improved the operation of such systems by the reduction of both the undershoot, caused by a non-minimum phase, and other oscillations caused by bad damping. The presented method improves the system dynamics and is suitable for real-time control, because its simplicity after the predictive coefficients are correctly defined. For open loop stable systems, we have shown that the control of such a system with improved MPC-algorithm is stable under defined conditions. Genetic algorithms have been used as search and as optimization method.

Keywords: predictive control, non-parametric control, non-minimum phase, optimization, genetic algorithms, predictive coefficients

[full-paper]