Hybrid Deep Learning Model Proposed for Electric Vehicle Range Prediction

Com reported. The adequate forecast of the electric vehicle (EV) range is the key to the success of the improved energy management and the improved battery utilization, the abstract states. Existing methods have struggled to capture the nonlinear, dynamic and non-stationary characteristics of EV operation and driving data, resulting in limited prediction accuracy and applicability across vehicle classes and driving scenarios.

The authors proposed a novel hybrid deep learning architecture to address these challenges through high-level feature extraction, adaptive sequence modeling and attention weight optimization. The framework consists of three components: the Hilbert-Huang Auto Recurrence Transform (HART), the Path-Informed Adaptive Inverted Trend Hub Range Analyser (PAITHRA) and the Chaotic Algae Sparrow Hyper-tuner (CASH).

HART performs rich temporal and recurrent features extractions.

PAITHRA is a combination of predictive model using pathformer and adaptive routing of features. CASH is a dynamic attention weight optimizer.

It was tested on the EV Energy Consumption Dataset and the Full Electric Vehicle Dataset 2024, which contain different driving styles, battery conditions, and weather conditions. 991. It required 35 minutes total for all epochs up to convergence and maintained an inference time of 8 ms per sample.

The proposed model is superior to conventional deep learning models. The paper's keywords are: Electric vehicles (EVs), Driving range prediction, Deep learning, Energy consumption modeling, Feature extraction, And optimization. Authors are Belqasem Aljafari, Thanikanti Sudhakar Babu, and Shitharth Selvarajan.

Belqasem Aljafari is affiliated with the Electrical Engineering Department, College of Engineering, Najran University, Najran, 11001, Saudi Arabia. Thanikanti Sudhakar Babu is affiliated with the Department of Electrical and Electronics Engineering, Chaitanya Bharati Institute of Technology, Telangana, India.

Shitharth Selvarajan is affiliated with the Department of CSE, University of Kebri Dehar, Kebri Dehar, Ethiopia and the Centre for Research Impact and Outcome, Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, Punjab, India.

The work was funded by the Deanship of Graduate Studies and Scientific Research at Najran University under the Elite Funding Program grant code NU/EP/SERC/14/2264-1. The authors declare no competing interests.

The paper was received on 08 December 2025 and accepted on 28 January 2026. 1038/s41598-026-37995-w. 0 International License.