Abstract:

Aiming at the problems of limited selection range of voltage vector and low accuracy of current tracking in the classical model predictive current control algorithm, an improved model predictive voltage control algorithm was proposed. The better voltage vector was obtained by rolling optimization, and using T_s/2 for the better voltage vector and T_s/2 for zero vector to achieve the half amplitude vector. The optimal voltage vector was chosen according to the evaluation function of the two vectors. Adding only one voltage vector per period could effectively reduce the error between predicted voltage vector and actual voltage vector, realized the accurate tracking of harmonic current, and improved the compensation performance of active power filter. Simulation results showed the proposed improved model predictive voltage control algorithm is more accurate than the classical model predictive current control algorithm in tracking the reference current. After compensation, the harmonic content in the grid-connected current decreases from 4.86% to 3.32%, and the grid-connected unit power factor can be achieved under both steady-state and dynamic conditions.

Key words: model predictive control, active power filter, current harmonics, optimal voltage vector