Researchers Build Electric Drive Digital Twin with Real-Time Load Compensation
New digital twin technique for PMSM motors virtually senses load torque via observer controller...
A team led by Frank Liebich has published a paper detailing the practical implementation of a digital twin (DT) for an electric drive in a networked test lab environment. The system centers on a permanent magnet synchronous machine (PMSM), a common motor in industrial and automotive applications. Unlike simpler monitoring concepts, this DT maintains full bidirectional data exchange between the physical motor and its digital counterpart, enabling not just observation but active control.
The key innovation is the integration of an observer controller into the PMSM control loop. This controller uses the digital twin to virtually sense the acting load torque—the disturbance variable—which is typically difficult to measure directly in real-world installations. The DT then adjusts the motor's operation to compensate for that disturbance, improving performance under varying loads. Test bench results confirm effective compensation, but the observed torque still deviates slightly from the physically measured torque, signaling that the model needs further calibration.
The work, accepted for the AmEC 2026 conference, is positioned as a first step toward broader practical deployment of digital twins in electric drive testing. Future refinements could reduce estimation errors and extend the approach to other motor types or more complex drivetrains. For engineers working on motor control and predictive maintenance, this offers a tangible path to virtual sensing and closed-loop DTs that improve real-world hardware.
- Implements a digital twin with bidirectional data exchange for a PMSM electric drive, enabling automated control alongside monitoring.
- Uses an observer controller within the PMSM control loop to virtually sense and compensate for load torque, a disturbance variable often unmeasurable directly.
- Test bench results demonstrate effective compensation but show a deviation between observed and measured torque, indicating room for model refinement.
Why It Matters
Enables virtual sensing of load torque, reducing need for physical sensors and improving electric drive control in real-world applications.