Particle swarm optimization of the multioscillatory LQR for a three-phase grid-tie converter

This paper presents an evolutionary optimization of the linear-quadratic (LQ) current controller for a three-phase grid-tie voltage source converter with an L-type input filter. The current control system is equipped with multi-oscillatory terms, which enable the converter to obtain nearly sinusoidal shape and balanced input currents under unbalanced and distorted grid voltage conditions. The augmentation of the state vector to include additional states which describe dynamics of disturbances increases the number of weights to be selected for a cost function in the LQR procedure design. Therefore, it is proposed that optimal weighting factors are sought using particle-swarm-based method. Finally, the simulational tuning based on the linear model and the numerical verification based on a non-linear model of the system with a pulse width modulator are addressed.