Predictive Modelling of Battery Behaviour for Enhanced Energy Management in Electric Vehicles Altair Engineering India, Pvt., Limited

Format:

Book

Conference/Event

Author/Creator:

G, Ayana, author.

Contributor:

Gumma, Muralidhar

Conference Name:

Symposium on International Automotive Technology (2026) (2026-01-28 : Pune, India)

Language:

English

Physical Description:

1 online resource cm

Place of Publication:

Warrendale, PA SAE International 2026

Summary:

The transition to electric vehicles is a significant change as the world moves toward sustainable objectives, and thus the effective usage of energy and batter functioning. However, accurate battery modelling and monitoring is still challenging due to its highly nonlinear behaviour because of its dependencies with temperature variations, aging effects, and variable load conditions.To address these complexities, there are smart battery management systems that monitor the key parameters like voltage, current, temperature, and State of Charge, ensuring safe and efficient battery operation. At the same time, this may not completely capture the battery's dynamic aging behaviour. Here, digital twin emerges as the powerful solution, which replicates the complete physical system into a virtual platform where we can monitor, predict and control.This research paper shows the digital twin solution framework developed for the real-time monitoring and prediction of key battery parameters and faults. An integrated modelling approach which combines a physics-based model which couples electrical, thermal and aging behaviour of the battery together with machine learning model for performance parameters and fault prediction makes the solution reliable and adaptable. Capacity fade and internal resistance growth is captured through the degradation factor that evolves with charge-discharge cycles, allowing the model to account for battery aging.The proposed digital twin framework tackles the challenges of nonlinear battery behaviour by combining physics-based and machine learning based modelling approach, with minimal real-time sensor data showed the results within an error margin of 3%. By testing and validating the model under real world driving conditions and loads strengthened the system reliability and adaptability

Notes:

Vendor supplied data

Publisher Number:

2026-26-0385

Access Restriction:

Restricted for use by site license