Adaptive Deadbeat Predictive Controller for Grid-feeding Inverter with Online Filter Inductance Estimation using Recursive Least Squares

This work proposes an adaptive predictive control scheme with online parameters identification to mitigate the inductive filter parameter mismatch in grid-connected inverters. The effect of the parametric mismatches on the current prediction error and the stability of the closed loop control system is investigated. The predictive current controller is implemented in a rotary reference frame following the voltage-oriented control (VOC) principle. The dc components of the grid current in the d-q reference frame during steady-state allowing an easy implementation of the recursive least squares estimator in order to estimate the filter inductance and resistance parameters. The estimated parameters online are transmitted to the predictive current controller to accurately calculate the output predictions and generate the optimal control action. As a result, the inherent sensibility of the predictive controller to the parameters mismatch is avoided. The effect of the weighting factor of the recursive least squares estimator on the controller performance is discussed.