EV Energy Consumption Digital Twin and Range Prediction

How far can an electric car still drive with the energy left in its battery - on a cold morning, on the highway, with the heater on? This project builds a digital twin of three real Nissan Leaf electric cars from their own second-by-second driving logs in the open Vehicle Energy Dataset (VED, Ann Arbor, Michigan, 2017-11 .. 2018-11). A longitudinal vehicle-dynamics model computes the force at the wheels (inertia, aerodynamic drag, rolling resistance, road grade from a terrain model), turns it into battery power with drive and regeneration efficiencies and adds the auxiliary and heater / A/C power. Its parameters are calibrated per car, and a gradient-boosting model of the remaining error makes it a hybrid twin. The twin predicts the energy per km of unseen trips and the remaining range as a function of outside temperature, HVAC use and driving pattern, compared honestly with the nominal physics model, pure machine-learning models and the EPA rating. A web application on five databases runs the twin on real or uploaded trips and plans the range.

Category: Automobile Engineering

Automobile Engineering Project AU03 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.