Real-Time Grid Impedance Identification for Online Parameter Tuning of Predictive Control in Grid-Tied Converters Using Artificial Neural Networks

With Europe's ambitious goals of reducing greenhouse gas emissions, the gradual replacement of traditional fossil-based generators by grid-connected converters is becoming prominent. Synchronization stability plays a crucial role in the operation of grid-forming converters, where power grid characteristics, particularly grid impedances, significantly impact synchronization stability. This paper presents a novel approach to grid impedance identification using the extended Kalman filter (EKF) concept. The identified parameters are utilized to fine-tune the model-predictive controller (MPC) of the converter's current control loop, ensuring strict constraints on input and output variable amplitudes. The performance of the proposed algorithm heavily relies on the weighting factors employed in both the MPC and EKF algorithms, posing a challenge in the existing literature. To address this issue, an artificial neural network (ANN) is employed to optimize the weighting factors in the proposed algorithm. Experimental tests conducted in PowerLabDK are presented to evaluate the effectiveness and performance of the proposed approach. The results demonstrate an accuracy of over ninety-six percent in the identified parameters during the experimental tests.