Diagnostics of Hybrid Electrical Machines in Electromechanical Systems

Abstract

The Hybrid Electrical Machine (HEM) is a key component of modern electromechanical drive systems, combining engine cranking and electrical power generation within a single compact device. Reliable HEM operation directly affects system dependability and energy efficiency. Faults in machine windings, power electronics, or sensor subsystems can cause performance degradation, downtime, or catastrophic failure. This paper presents a diagnostic approach for a Permanent Magnet Synchronous Starter Generator (PMSSG) operating in a variable-speed electromechanical system. The proposed method uses three measurable quantities: stator phase current, terminal voltage, and rotor angular speed. Unlike computationally intensive spectral or machine-learning methods, the proposed approach relies on deviations of these signals from their healthy reference trajectories, particularly during transients, where fault signatures are most pronounced. A main contribution is the Integrated Fault Index (IFI), defined as a weighted Euclidean norm of normalized signal deviations, enabling continuous real-time assessment and fault detection without historical training data or supervised learning procedures. A Tabular–Logical Diagnostic Model (TLDM) is also proposed. The TLDM associates characteristic signal patterns with specific fault classes, including winding inter-turn short circuits, inverter switch failures, sensor offset faults, overload conditions, and rotor eccentricity. The diagnostic framework is implemented and validated in MATLAB/Simulink. Simulation results show high fault detection sensitivity in all HEM operating modes, low false-alarm rates during normal transients, and robust performance under parameter uncertainty. Comparative analysis with conventional single-signal and FFT-based methods confirms the improved diagnostic coverage of the proposed approach.