New Fault Diagnosis Method Enhances Reliability of EV Motors
In the rapidly evolving world of electric vehicle (EV) technology, where performance, efficiency, and safety are paramount, researchers continue to push the boundaries of motor design and fault detection. A recent breakthrough in motor diagnostics, published in the Proceedings of the CSEE, introduces a novel method for detecting internal winding faults in a specialized type of permanent magnet motor commonly considered for next-generation EVs. The work, led by Zhenfei Chen and her team from Hohai University, presents a smart, cost-effective solution to a long-standing challenge in motor reliability—detecting open-circuit faults in inaccessible internal windings.
As automakers strive to deliver vehicles with smoother performance, longer range, and higher durability, the electric motor remains a core component under constant scrutiny. Among the various motor topologies, fractional-slot permanent magnet machines (FS-PMMs) have gained attention for their high power density, compact design, and reduced torque ripple—key attributes for premium driving experiences. However, one variant, the star-delta (Y-Δ) hybrid-connected FS-PMM, while offering superior harmonic suppression and lower losses, comes with a hidden vulnerability: its inner delta-connected windings are physically isolated and not directly accessible for current measurement.
This architectural feature, while beneficial for electromagnetic performance, creates a blind spot for fault monitoring. When a single phase in the inner delta winding experiences an open-circuit fault—essentially a broken wire—there is no immediate indication at the motor terminals, where sensors are typically placed. Unlike faults in the outer star-connected windings, which cause an obvious drop in measured phase current, internal delta faults can go unnoticed during early stages. Left undetected, such faults can lead to increased thermal stress, unbalanced magnetic forces, and ultimately, catastrophic motor failure. This poses a significant risk, particularly in safety-critical applications like electric propulsion systems.
Conventional fault diagnosis methods often rely on measuring zero-sequence currents or neutral-point voltages—signals that are only present in star-connected systems. These techniques are ineffective for delta-connected or hybrid Y-Δ configurations. Other approaches, such as model-based predictive control or dq-axis harmonic analysis, require complex control modifications and additional computational overhead, making them less attractive for real-time, low-cost implementations. The lack of a reliable, non-intrusive diagnostic method for internal delta faults has been a notable gap in the field.
Chen Zhenfei, an associate professor at Hohai University’s School of Electrical and Power Engineering, recognized this challenge and led a team to develop a new diagnostic strategy that leverages inherent motor characteristics rather than requiring new hardware. Their approach, detailed in the November 2024 issue of the Proceedings of the CSEE, focuses on the third harmonic component of the stator current—a signal often considered noise but, in this case, becomes a powerful diagnostic indicator.
The insight behind the method lies in understanding the electromagnetic behavior of the Y-Δ hybrid system under fault conditions. In normal operation, the symmetrical three-phase currents in both the star and delta windings cancel out triplen harmonics (such as the third, ninth, etc.), resulting in negligible third harmonic content in the measurable outer phase currents. However, when a single phase in the inner delta winding opens, this symmetry is disrupted. The broken circuit prevents the normal circulation of third harmonic currents within the delta loop, forcing these harmonics to seek an alternative path through the outer star windings.
As a result, the phase currents in the accessible star-connected winding exhibit a measurable increase in third harmonic amplitude. Crucially, the distribution of this harmonic is not uniform across all three phases. The team’s analysis revealed a distinct pattern: in the event of an X-phase open circuit, the A and C phase currents show elevated third harmonic levels, while the B phase remains relatively unaffected. Similarly, a Y-phase fault elevates harmonics in A and B, leaving C clean, and a Z-phase fault impacts B and C, sparing A. This unique “signature” allows not only for fault detection but also for precise identification of the faulty phase.
While this harmonic-based approach is elegant in theory, real-world implementation faces practical hurdles, particularly at low motor speeds. At low RPMs, the magnitude of the third harmonic current is inherently small, often comparable to the noise floor introduced by current sensors and signal processing circuits. In such conditions, distinguishing a genuine fault signal from measurement error becomes challenging, potentially leading to false alarms or missed detections.
To overcome this limitation, Chen and her colleagues introduced a two-step signal enhancement process. First, they apply a squaring operation to the extracted third harmonic amplitudes. This non-linear transformation amplifies the differences between phases with high harmonics and those with low, making the fault signature more pronounced. Second, they scale the squared values by a gain factor, effectively boosting the signal-to-noise ratio. This amplified signal is then compared against a predefined threshold to generate a binary “fault phase location index.”
The index is designed for clarity and reliability. In healthy operation, all three phase indices are zero. When a fault occurs, two indices become “1” while the one corresponding to the least affected phase remains “0.” For example, if the B-phase index stays at zero while A and C rise to one, the system identifies an X-phase fault in the inner delta winding. This logical mapping enables both rapid fault detection and accurate phase localization with minimal computational effort.
The robustness of the method was rigorously tested on a physical prototype of a 10-pole, 12-slot Y-Δ hybrid FS-PMM. The experimental setup included a programmable load motor, a high-precision torque sensor, and a real-time control system based on RTUBox, allowing the researchers to simulate open-circuit faults by remotely opening circuit breakers in the inner winding connections. Tests were conducted across a wide range of operating conditions, including varying speeds from standstill to 800 RPM and dynamic load changes.
Results confirmed the theoretical predictions. Under normal conditions, the third harmonic content in the outer phase currents was negligible. Upon inducing an open-circuit fault in the X-phase of the inner delta winding, the third harmonic amplitudes in the A and C phases of the outer star winding immediately increased, while the B-phase harmonic remained near zero. The fault phase location index responded instantaneously, switching from all zeros to the correct “1-0-1” pattern, clearly indicating an X-phase fault.
Perhaps most impressively, the method demonstrated exceptional performance at low speeds, where traditional approaches often fail. Even at 200 RPM, the amplified harmonic signal clearly distinguished the faulty condition from noise, thanks to the squaring and scaling technique. The team carefully selected the amplification factor and threshold to balance sensitivity and reliability, ensuring that minor fluctuations due to sensor noise would not trigger false alarms. Additional tests with sudden load changes—simulating real-world driving conditions like rapid acceleration or hill climbing—showed that the diagnostic index remained stable and accurate, unaffected by transient torque demands.
One of the most compelling advantages of this new method is its practicality. It requires no additional sensors, no modification to the motor’s physical structure, and no access to the internal windings. It operates solely on the standard phase current measurements already available in most motor control systems. This makes it an ideal candidate for integration into existing EV motor controllers, adding a layer of fault protection without increasing system cost or complexity.
From a manufacturing and maintenance perspective, this technology offers significant value. Early detection of winding faults allows for predictive maintenance, preventing unexpected breakdowns and reducing long-term ownership costs. For automakers, it enhances the perceived reliability and safety of their EVs, a key selling point in a competitive market. The method also supports the broader adoption of Y-Δ hybrid motors, which, despite their performance benefits, have been underutilized due to diagnostic challenges.
The implications extend beyond passenger vehicles. This diagnostic technique could be applied to any system using Y-Δ hybrid-connected motors, including electric buses, commercial trucks, industrial drives, and even aerospace applications where motor reliability is critical. The principle of using inherent harmonic content as a diagnostic signal could inspire similar approaches for other types of electrical machines and fault conditions.
Chen Zhenfei’s work represents a shift from reactive to proactive motor health monitoring. Instead of waiting for a fault to escalate into a major failure, the system can identify the problem at its inception, allowing for timely intervention. This aligns with the growing trend in the automotive industry toward predictive analytics and digital twins, where vehicle systems continuously monitor their own health and performance.
The research also highlights the importance of interdisciplinary thinking in engineering innovation. By combining deep knowledge of motor electromagnetic theory with practical signal processing techniques, the team was able to turn a potential weakness—the presence of harmonics—into a powerful diagnostic tool. This kind of creative problem-solving is essential for advancing the state of the art in electric mobility.
As EVs become more complex, with multiple motors, advanced power electronics, and sophisticated control algorithms, the need for intelligent, embedded diagnostics will only grow. Methods like the one developed by Chen and her team provide a blueprint for building smarter, more resilient electric drive systems. They demonstrate that sometimes, the most effective solutions are not the most complex, but rather those that cleverly leverage the natural behavior of the system.
Looking ahead, this diagnostic framework could be expanded to detect other types of faults, such as short circuits or partial winding degradation, by analyzing different harmonic components or using machine learning to identify more subtle patterns. Integration with vehicle telematics systems could enable remote monitoring of motor health, providing fleet operators with real-time insights into the condition of their vehicles.
In conclusion, the research conducted by Zhenfei Chen, Feng Wang, Zhixin Li, Xiangmin Wan, and Zhihao Ling at Hohai University marks a significant step forward in the reliability and safety of electric motors for automotive applications. Their harmonic-based fault diagnosis method for Y-Δ hybrid FS-PMMs is not only scientifically sound but also eminently practical, offering a robust, low-cost solution to a persistent engineering challenge. As the automotive industry continues its electrification journey, innovations like this will play a crucial role in ensuring that electric vehicles are not only efficient and powerful but also dependable and safe.
Zhenfei Chen, Feng Wang, Zhixin Li, Xiangmin Wan, Zhihao Ling, Hohai University, Proceedings of the CSEE, DOI: 10.13334/j.0258-8013.pcsee.231539