In-Depth Performance Evaluation of Existing Sensorless Control Methods for Permanent Magnet Synchronous Motors in Electric Vehicles

Abstract

Permanent Magnet Synchronous Motors (PMSMs) are critical components widely utilized in high-performance applications such as electric vehicles, industrial automation, robotics, and renewable energy systems due to their high efficiency, power density, torque density, and reliability. However, the requirement for mechanical position sensors increases system cost and volume while simultaneously reducing overall reliability. Sensorless control strategies aim to eliminate these limitations by accurately estimating the rotor position and speed using only electrical variables, thereby reducing system cost and enhancing reliability. This paper presents a comprehensive performance evaluation of recent and classical sensorless control strategies designed to operate across the entire speed range. The algorithms are broadly categorized into model-based and saliency-based approaches. The in depth analysis drawing on recent and influential studies, specifically highlights compound, hybrid, and predictive frameworks, covering influential techniques such as back-EMF estimation, sliding-mode observers (SMO), model reference adaptive systems (MRAS), extended Kalman filters (EKF), phase-locked loops (PLL), and high-frequency signal injection (HFI). The study compares over entire speed ranges a critical performance metrics, including estimation accuracy, dynamic performance, computational efficiency, torque ripple, noise immunity, complexity and cost. Recent trends discussed include the adoption of adaptive observers and advanced artificial-intelligence (AI) estimators for improved position estimation. To sum up, the integration of hybrid sensorless control with data-driven learning and predictive optimization offers the most promising path for creating automotive-grade PMSM drives that are affordable, compact, and dependable, while also providing accurate position estimation in all operating scenarios. Despite advancements, the paper summarizes ongoing challenges in the field, specifically noting issues related to parameter sensitivity, computational cost, and ensuring smooth transition between different operating modes. Prospects for future research are also provided based on the comprehensive analysis.