EVE – Extending Life of Vehicles within Electromobility Era

Hybrid and fully electric buses must take advantage of predictive maintenance services based on Machine Learning (ML) in order to remain competitive in the market despite their increased cost. Those new technical solutions are enabled by the possibility of collecting more precise and extensive data. Motivating example is the main battery, a component that is responsible for large fraction of the vehicle cost, and is expected to survive over 10 years.

In the project we will create a data-driven model for battery health degradation, based on Transfer Learning (TL) paradigm utilising recent advancements in the field of deep learning, namely Generative Adversarial Networks (GANs). Historical data on failures of electric driveline components is not available due to its technological novelty, thus the right support for drivers, fleet operators and OEMs requires studies and development of new ML techniques.

Our goal is to enable predictive and prescriptive maintenance solutions that are capable of adapting to rapid changes in technology and continuous arrival of new observational data. This must be bootstrapped by leveraging available expert knowledge in the form of adaptive models for the evolving technology through AI solutions.