Adaptive Neural Network Control for Sensorless DC Motor Speed Regulation in Electric Vehicles under Road Disturbances
Abstract Regulating the speed of an electric vehicle (EV) drive is hard: the dynamics are nonlinear, the motor parameters are never known exactly, and the road keeps changing the load. This paper presents a disturbance-aware adaptive sensorless neural network controller (DAAS-NNC) for a DC motor drive in an EV. A neural network estimates the disturbance online, and an adaptive learning law driven by the error derivative tunes the controller, improving the transient response and robustness without affecting steady-state accuracy. The controller needs no mechanical speed sensor, which lowers hardware cost and removes a component that can fail. A Lyapunov-based adaptive law keeps the closed-loop system stable and all signals bounded. The controller is tested for speed tracking, load variation, and sloped and banked road profiles. In simulation, DAAS-NNC reduces the rise time by 44.4% and 21.9% and the settling time by 55.6% and 33.3% relative to conventional PI and MRAS controllers, respectively, and cuts the overshoot from 12.3% (PI) and 5.6% (MRAS) to 1.8%. It also reduces speed deviations, produces smoother control inputs with fewer oscillations, and lets the neural network estimator converge quickly. These results point to DAAS-NNC as a promising controller for EV powertrains.
Authors
- Vipin Chandra Pal (ORCID: https://orcid.org/0000-0003-2750-8528)
- Sheetla Prasad (ORCID: https://orcid.org/0000-0001-6517-5002)
- Akshaya Kumar Pati (ORCID: https://orcid.org/0000-0003-3877-1575)
- Nitendra kumar
Institutions
- National Institute Of Technology Silchar (IN)
- Galgotias University (IN)
- KIIT University (IN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-13
- DOI
- https://doi.org/10.21203/rs.3.rs-11354505/v1
- Primary Topic
- Sensorless Control of Electric Motors
- Type
- preprint