Machine Learning-based Trajectory Planning for Single-loop Flatness-based Control of PMSMs

Permanent magnet synchronous motors (PMSMs) are among the most widely used motors in modern industry. Over the past few decades, extensive research has been conducted on various control methods, while field-oriented control (FOC) being one of the most well-known approaches. Additionally, flatness-based (FB) control has been introduced in the literature as a solution for addressing the nonlinear characteristics inherent in PMSM drive systems. Traditional FB control methods typically has a cascaded structure and employ second-order functions as trajectory functions to maintain the flatness property of the drive system. However, this cascaded structure presents certain limitations, particularly in applications requiring high-dynamic performance. To overcome these drawbacks, the concept of single-loop FB control has been proposed in recent studies. One significant challenge in single-loop FB control systems is ignoring controller limits, such as overcurrent protection. To address this issue, this paper proposes a novel trajectory planning method for single-loop FB control of PMSMs, using machine learning. The proposed method effectively tackles the challenge of current protection while maintaining the system’s flatness property. The effectiveness of the proposed approach has been validated through simulation studies, demonstrating its potential for enhancing the performance of PMSM drives.