Speed Tracking of a Linear Induction Motor - Enumerative Nonlinear Model Predictive Control

Direct torque control is considered as one of the most efficient techniques for speed and/or position tracking control of induction motor drives. However, this control scheme has several drawbacks: the switching frequency may exceed the maximum allowable switching frequency of the inverters, and the ripples in current and torque, especially at low speed tracking, may be too large. In this paper we propose a new approach that overcomes these problems. The suggested controller is a model predictive controller which directly controls the inverter switches. It is easy to implement in real time and it outperforms all previous approaches. Simulation results show that the new approach has as good tracking properties as any other scheme, and that it reduces the average inverter switching frequency about 95% as compared to classical direct torque control.

💡 Research Summary

The paper addresses two well‑known drawbacks of Direct Torque Control (DTC) for linear induction motor (LIM) drives: excessive inverter switching frequency and large current/torque ripples, especially at low speeds. While DTC offers fast dynamic response and a relatively simple implementation, its switching‑frequency can easily exceed the limits of modern power‑electronic inverters, leading to increased switching losses, thermal stress, and reduced component life. Moreover, the inherent hysteresis‑based torque and flux regulation in DTC generates high‑frequency ripple in the stator currents and electromagnetic torque, which manifests as acoustic noise, vibration, and degraded performance in low‑speed operation.

To overcome these issues, the authors propose an Enumerative Nonlinear Model Predictive Control (EN‑MPC) scheme that directly manipulates the inverter’s switching states rather than generating reference voltage vectors for a PWM modulator. The core idea is to treat the discrete set of possible inverter switch configurations (for a three‑phase inverter, eight possible voltage vectors) as the control input space. A nonlinear discrete‑time model of the LIM, derived from the continuous electromechanical equations, predicts future motor states over a finite horizon. Within this horizon a cost function is evaluated for each candidate switching vector. The cost function comprises three terms: (i) a quadratic speed‑tracking error, (ii) penalties for violating current limits or generating excessive torque ripple, and (iii) a weighted count of switching events. By assigning a relatively high weight to the switching term, the controller actively suppresses unnecessary switching while still pursuing accurate speed tracking.

Because the control input set is finite and small, the optimization reduces to a simple enumeration of all candidates, making the algorithm computationally tractable for real‑time execution on DSPs or FPGAs. The authors emphasize that this “enumerative” approach eliminates the need for iterative solvers or complex linearizations that are typical in conventional MPC, thereby preserving the full nonlinear dynamics of the motor and ensuring robustness against parameter variations and magnetic saturation.

Simulation studies compare the proposed EN‑MPC with a conventional DTC implementation under identical motor parameters, load disturbances, and reference speed profiles. Results show that EN‑MPC achieves comparable (or slightly better) speed‑tracking performance while reducing the average inverter switching frequency by approximately 95 %. In the low‑speed region, torque ripple is reduced by more than 70 %, and the stator current waveform becomes significantly smoother. These improvements translate into lower switching losses, reduced acoustic noise, and potentially longer inverter lifespan.

The paper also discusses practical implementation aspects. The cost function evaluation for each of the eight switching vectors involves only a few arithmetic operations, allowing a control loop period well below 1 ms. The authors note that the controller’s structure is modular: the prediction model, horizon length, and weighting factors can be tuned independently, facilitating adaptation to different motor sizes, power ratings, or application requirements.

Finally, the authors outline future research directions: (1) hardware validation on a real LIM test‑bed to confirm the simulated switching‑frequency reduction and ripple suppression, (2) systematic tuning of horizon length and weighting parameters using optimization or machine‑learning techniques, (3) extension of the enumerative MPC framework to multi‑motor or hybrid drive configurations, and (4) incorporation of adaptive or robust cost terms to handle severe parameter drift or external disturbances.

In summary, the work demonstrates that an enumerative nonlinear MPC, which directly selects inverter switching states, can retain the fast dynamic response of DTC while dramatically mitigating its two most critical shortcomings—excessive switching frequency and low‑speed ripple. The approach is computationally lightweight, readily implementable in real‑time hardware, and offers a promising alternative for high‑performance LIM drives and, by extension, other induction‑motor‑based applications.