Project Description

The number of electric vehicles are expected to increase over the years increasing the dynamic and mobile loads connected to the power systems, creating various operational challenges for the electric grids. This Ph.D. project deals with the development of a dual-layer Digital Twin framework that utilises high speed edge computing for real-time state tracking and cloud-based "Big Data" analytics for longterm degradation modelling, specifically addressing the non linear ageing of Lithium-ion batteries.

Current Battery Management Systems (BMS) often rely on Coulomb Counting or simple Equivalent Circuit Models (ECM), which fail as batteries age or operate under extreme temperatures. There is a scope for Physically Informed Machine Learning (PIML) that respects electrochemical laws while benefiting from AI's computational power, which may be known as the Physics Informed AI Engine. The possibility of reduced order modelling can be explored for the fast calculations. Integration of the light weight AI based models on the electric vehicles BMS would help to give the signals for the health monitoring and the control signals for the dynamic safe operating regions based on the various temperature for the aird environments.

Further, the created digital twin would be utilised for various power system studies, including scheduling of electric vehicle charging/discharging while their participation in G2V and V2G operations. The grid operator should have information on the health cost metric of a vehicle during scheduling to support functions such as frequency control and peak shaving, while considering the health of EV batteries and minimising overall battery degradation.