A Novel Approach to Performance Evaluation of Current Controllers in Power Converters and Electric Drives Using Non-Parametric Analysis

Abstract

In the field of current controllers for power converters and electric motor units, conventional figures of merit (FMs) such as mean squared error (MSE), integral of time multiplied by absolute error (ITAE), or total harmonic distortion (THD) have traditionally been used. The stochastic nature of these FMs introduces variability, which often results in inaccurate comparisons of the controllers' performance. To address this issue, a parametric statistical methodology has recently been proposed. However, it presents certain limitations, such as the assumption of normality. In response to this, the adoption of a non-parametric methodology is suggested in this work, which promises a precise evaluation of the efficiency of current controllers. In this study, we demonstrate that, under parametric approach, when the assumption of normality is violated, there is a significant increase in Type I error. Furthermore, we show that the non-parametric Mann-Whitney U test offers greater sensitivity compared to its parametric counterpart under these circumstances. Thus, the newly proposed methodology aims to optimize the decision-making process in designing high-performance current controllers for applications in power converters and electric motors. This allows for design decisions grounded in rigorous and statistically-based evaluations. The effectiveness of this methodology is confirmed through its application to a real dataset, enhancing its practicality and contributing to a deeper understanding of the subject matter.