Context-Aware Battery Health Modelling Using an Entangled Variational Quantum Model

Lithium-ion (Li-ion) battery degradation under Electric Vehicle (EV) operating conditions remains under-explored, as most State of Health (SoH) prediction models rely on laboratory datasets that omit operational context such as climate and charging behaviour. This creates a gap between laboratory-trained models and real-world EV deployment. In this paper, we present a context-aware entangled Variational Quantum Neural Network (VQNN) for the early classification and prediction of battery failures. Agent-Based Modelling (ABM) is used to simulate the operational context, which is then integrated with discharge data from NASA battery datasets B0005 and B0006. The performance of the entangled VQNN is compared with that of the non-entangled VQNN and classical Machine Learning (ML) techniques. The results show that operational context affects battery degradation, where hot climates and high charging rates reduce battery lifespan. Our findings show that the entangled VQNN achieves a superior performance, with an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.95, an F1-score of 0.82, and an accuracy of 88.52%. It outperforms baseline models such as Random Forests, Decision Trees, Support Vector Machines (SVM) and non-entangled VQNN which achieve AUC-ROC scores of 0.94, 0.93, 0.90 and 0.92, respectively. We conclude that entangled quantum models improve the representation of latent contextual interactions, which translates into enhanced predictive performance and supports the development of more robust Battery Management Systems (BMS).