TY  - JOUR
AU  - Zhao, Leiting
AU  - Ruan, Zheng
AU  - Liu, Kan
AU  - Li, Liran
AU  - Zou, Yuchao
PY  - 2025
TI  - A digital-twin-based open circuit fault diagnosis method for permanent magnet motor drive system
JO  - Railway Sciences
SN  - 2755-0907
SP  - 494
EP  - 521
VL  - 4
IS  - 4
AB  - PurposeThis study aims to implement condition monitoring for urban rail train permanent magnet synchronous motors and inverter systems. Through the construction of a digital twin model, it performs fault diagnosis of potential system failures, enabling rapid fault localization and protection.Design/methodology/approachThis research begins with a brief introduction to the structure and classification of permanent magnet synchronous motors (PMSMs), followed by a detailed analysis of their mathematical model. Subsequently, it thoroughly investigates the working principle of three-phase two-level inverters and the distribution of space voltage vectors. Based on the analysis of the main circuit topology, a digital twin model matching the external characteristics of the physical circuit is established using the model predictive control method, achieving accurate system simulation. Furthermore, through theoretical analysis and simulation verification of phase current characteristics under inverter switch tube faults, general patterns of phase currents under fault conditions are summarized. The established digital twin model is then employed to validate these patterns, confirming the model’s effectiveness in fault diagnosis.FindingsThis study proposes a fault diagnosis method based on digital twins. Experimental and simulation results demonstrate that the established digital twin model can accurately simulate the external characteristics of the actual physical circuit, validating its effectiveness in inverter fault diagnosis. This approach offers practical value for condition monitoring in actual urban rail train systems.Originality/valueThe study innovatively starts from a mathematical model and simulates the actual physical model through a virtual model, requiring only external characteristics to achieve system fault diagnosis, thereby enhancing diagnostic efficiency.
UR  - https://doi.org/10.1108/RS-04-2025-0008
DO  - 10.1108/RS-04-2025-0008