Advanced Control Strategy Boosts Dual Three-Phase PMSM Performance in EVs
In the rapidly evolving landscape of electric mobility, the quest for more efficient, reliable, and robust electric drive systems has never been more critical. As automakers and technology developers race to refine their powertrains for next-generation vehicles, a groundbreaking advancement in motor control methodology is capturing the attention of engineers and researchers worldwide. A newly published study introduces a novel robust model predictive control (MPC) strategy specifically tailored for dual three-phase permanent magnet synchronous motors (PMSMs)—a technology increasingly favored in high-performance and safety-critical applications such as electric vehicles (EVs) and aerospace systems.
At the heart of this innovation lies a sophisticated integration of sliding mode observer (SMO) techniques with traditional model predictive current control, enhanced by real-time parameter perturbation estimation and delay compensation. The result is a control architecture that not only accelerates dynamic response and minimizes current ripple but also dramatically improves system resilience against real-world uncertainties—such as variations in motor resistance, inductance, and magnetic flux—without compromising stability or efficiency.
The research, led by Changzheng Zhang, Yidan Ding, and Lei Yuan from the Hubei Collaborative Innovation Center for High-efficiency Utilization of Solar Energy at Hubei University of Technology, represents a significant leap forward in addressing one of the most persistent challenges in advanced motor control: sensitivity to parameter mismatch. In practical EV applications, motor parameters inevitably drift due to temperature fluctuations, aging components, or manufacturing tolerances. Traditional control strategies often falter under such conditions, leading to degraded performance, increased torque ripple, or even instability. This new approach directly confronts that vulnerability.
Dual three-phase PMSMs, which consist of two independent three-phase winding sets spatially offset by 30 electrical degrees, offer compelling advantages over conventional three-phase counterparts. These include reduced torque pulsation, lower current harmonics, enhanced fault tolerance (allowing continued operation even if one phase fails), and improved power density. However, harnessing these benefits requires equally advanced control algorithms capable of managing the system’s increased complexity. The authors tackle this by employing a dual d-q coordinate transformation—a mathematical framework that decouples the two winding sets into independent rotating reference frames, enabling precise and parallel control of each subsystem.
The core of their methodology builds upon finite control set model predictive control (FCS-MPC), a technique known for its simplicity, fast dynamic response, and ability to handle multi-objective optimization without the need for complex PI tuning. Yet, standard MPC is notoriously sensitive to inaccuracies in the motor model. Even minor deviations in stator resistance or inductance values can lead to significant prediction errors, causing suboptimal voltage vector selection and degraded current tracking.
To mitigate this, the team introduces an online parameter estimation mechanism via a sliding mode observer. Sliding mode control is renowned for its robustness against disturbances and model uncertainties, making it an ideal candidate for real-time compensation. The SMO continuously estimates the aggregate effect of parameter perturbations—such as ΔR (resistance deviation), ΔLd/ΔLq (inductance mismatches), and Δψf (flux linkage error)—and feeds these estimates back into the predictive model. This closed-loop correction ensures that the controller always operates with an effectively “updated” model, even as physical parameters shift during operation.
Furthermore, recognizing the inherent computational and switching delays in digital control systems—where the optimal voltage vector selected in one control cycle is only applied in the next—the researchers incorporate a two-step delay compensation scheme. By predicting the current state two sampling periods ahead (k+2), rather than just one (k+1), the controller anticipates and neutralizes the destabilizing effects of latency. This refinement is crucial for high-bandwidth applications like EV traction drives, where millisecond-level timing directly impacts ride quality and energy efficiency.
The team validated their SMO-MPC strategy through extensive simulations in MATLAB/Simulink, benchmarking it against conventional MPC under demanding scenarios. In one test, the motor was subjected to abrupt load torque changes—simulating real-world conditions like rapid acceleration or hill climbing. The SMO-MPC system demonstrated markedly superior performance: speed stabilized within 37 milliseconds after load steps, compared to 45 ms for traditional MPC, with significantly reduced overshoot (under 2% vs. 4.8%). More importantly, electromagnetic torque ripple—a key contributor to mechanical noise and vibration—was substantially suppressed.
In another critical test, the researchers deliberately introduced severe parameter perturbations. At various intervals, stator inductance was halved or doubled, resistance was varied between 50% and 200% of nominal values, and rotor flux linkage was adjusted by ±10%. Under these extreme conditions, the conventional MPC exhibited noticeable current distortion, tracking errors, and increased harmonic content. In stark contrast, the SMO-MPC maintained tight current regulation, with minimal steady-state error and remarkably low total harmonic distortion (THD). FFT analysis revealed a THD of just 1.03% for SMO-MPC, compared to 3.55% for standard MPC—translating to smoother torque delivery and quieter operation.
These improvements have profound implications for electric vehicle design. Lower current ripple means reduced copper losses and higher overall efficiency, extending driving range. Enhanced robustness against parameter drift allows for more relaxed manufacturing tolerances and eliminates the need for frequent recalibration, lowering production costs. Most critically, the improved fault resilience aligns perfectly with automotive safety standards (such as ISO 26262), where continued operability after partial system failure is not just desirable but often mandatory.
Moreover, the strategy’s compatibility with existing inverter hardware and digital signal processors makes it highly deployable. Unlike some advanced control methods that require specialized sensors or excessive computational resources, SMO-MPC operates entirely with standard current and voltage measurements and leverages algorithms well within the capability of modern automotive-grade microcontrollers.
The automotive industry is already witnessing a shift toward multi-phase drives in premium and performance EVs. Companies like Tesla, Lucid, and Rivian are exploring advanced motor topologies to extract maximum performance from their platforms. This research provides a ready-to-adopt control framework that could accelerate that transition, offering a clear path to higher efficiency, quieter operation, and greater reliability—all without increasing system complexity or cost.
Beyond electric vehicles, the implications extend to aerospace actuators, industrial robotics, and marine propulsion—any domain where power density, fault tolerance, and precision control are paramount. The dual three-phase PMSM, long considered a niche solution due to its control challenges, may now emerge as a mainstream alternative thanks to innovations like this.
The study also underscores a broader trend in control engineering: the convergence of classical robust control theory (like sliding mode) with modern predictive techniques. Rather than treating model uncertainty as a nuisance to be minimized through precise identification, this approach embraces it as an inevitable reality and designs controllers that actively compensate for it in real time. This paradigm shift is particularly well-suited to the unpredictable operating environments of mobile applications.
From a systems perspective, the integration of estimation and control into a unified framework represents a move toward more intelligent, adaptive power electronics. Future iterations could incorporate machine learning elements to further refine disturbance prediction or adapt observer gains based on operating conditions. But even in its current form, the SMO-MPC strategy delivers tangible, quantifiable benefits that address real engineering pain points.
As global regulations tighten on emissions and energy efficiency, and consumer expectations rise for EV performance and refinement, the pressure on powertrain developers intensifies. Breakthroughs like this—rooted in deep theoretical understanding yet focused on practical implementation—will be instrumental in bridging the gap between laboratory innovation and road-ready technology. The work by Zhang, Ding, and Yuan doesn’t just propose a better algorithm; it offers a more resilient, efficient, and ultimately more viable path forward for electric propulsion.
For automotive engineers, this research serves as both a technical blueprint and a strategic signal: the future of motor control lies not just in faster processors or more exotic materials, but in smarter, more adaptive algorithms that can thrive in the messy reality of the real world. And in that future, dual three-phase PMSMs, guided by robust predictive control, may well take the driver’s seat.
By Changzheng Zhang, Yidan Ding, and Lei Yuan, Hubei Collaborative Innovation Center for High-efficiency Utilization of Solar Energy, Hubei University of Technology. Published in Fire Control & Command Control, 2024, 49(8): 127–136. DOI: 10.3969/j.issn.1002-0640.2024.08.017.