Answers are grounded in this paper and cite their sources.
The transition to low-carbon sustainability in airport operations requires the deployment of high-performance, safe electric tow tractors. In this study, an improved predictive control method for electric towing vehicles (ETV) based on multi-step finite control set model predictive control (MS-FCS-MPC) with neural network (NN) accelerated cost function approximation is presented. The ETV powertrain examined models of a MULAG Comet 4E tow tractor, integrating a battery-powered induction-motor drive, a power converter, and a fixed mechanical transmission, while considering longitudinal vehicle dynamics. To address the concurrent objectives of ETV, accurate speed tracking, torque control, and effective battery operation and protection under high fluctuating tow loads, a multi-step prediction horizon is developed. A broader horizon avoids shortsighted control operations and energy inefficiency; however, it significantly increases real-time computational complexity due to the combinatorial explosion of switching states. To mitigate this issue, an offline-trained NN is included in the predictive control loop to estimate the cost function. This avoids exhaustive analytical cost evaluation in the online control loop while retaining the multi-step predictive formulation and providing a measured reduction in real-time decision latency. The fallback mechanism recalculates the predicted cost of the control signal vector analytically when significant deviations occur, thereby providing the system with high robustness. The proposed framework is assessed in two challenging test cases: a step-speed transient response test and a realistic airport towing mission based on the Aero driving cycle with dynamic speed fluctuations, regenerative braking, and varying slopes. Extensive numerical simulations in MATLAB/Simulink, followed by real-time controller (RTC) hardware-in-the-loop (HIL) testing using a dSPACE DS1104 controller, highlight the efficacy of the proposed framework The results demonstrate a measured reduction in real-time decision latency from 49.8 µs to 42.5 µs, corresponding to approximately 14.66%, while maintaining the evaluated tracking and energy-performance characteristics. Therefore, this strategy ensures safe, energy-efficient, and fast operation for airport operations.
No takes yet. Share an insight, caveat, or question.
Chiboub et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: