Online Filter Parameters Estimation in a Double Conversion UPS System for Real-Time Model Predictive Control Performance Optimization

Abstract

Uninterruptible Power Supplies (UPS) represent the key technology to continuously feed and protect a wide range of critical applications in electric power systems. Due to their continuous operation, UPS components degrade over time, including their filtering elements, leading to decreased filtering capabilities. With these deviations, UPS performance is typically reduced. Moreover, if filter parameter deviations are not considered in the control system, performance degradation can be further increased and the critical load can be seriously compromised. This is especially important with Model Predictive Control (MPC), which heavily relies on system parameters accuracy. Thus, control performance optimization using estimated filter parameters in UPS systems can be extremely important. Nevertheless, this has not yet been covered in the literature for UPS systems. In light of these facts, this paper proposes a new mechanism that, by using online estimated filter parameters, optimizes the performance of an MPC strategy, in a UPS system. In the scope of this optimization mechanism, an estimation method that enables parameter identification and control optimization not only in balanced but also in highly unbalanced filter conditions (rarely studied in the literature) is proposed. Experimental results demonstrate the accuracy of the proposed estimators and the effectiveness of the proposed control performance optimization mechanism. Under severe filter parameter variations, the proposed performance optimization scheme enabled to reduce the degradation (caused by filter variation) of the grid current and load voltage THD by 29.41% and 91.60%, respectively. Furthermore, the degradation of the RMS load voltage value was also significantly reduced by 97.89%