Real-Time Implementable Reduced-Order Energy Model for an Electric Vehicle

Accurate estimation of vehicle energy consumption plays an important role in developing advanced energy-saving connected automated vehicle technologies such as Eco Approach and Departure, PHEV mode blending, and Eco-route planning. The present study developed a reduced-order energy model with second-order response surfaces and torque estimation to estimate the energy consumption while just relying on the drive cycle information. The model is developed for fully electric Chevrolet Bolt using chassis dynamometer data. The dyno test data encompasses the various EPA test cycles, real-world, and aggressive maneuvers to capture most powertrain operating conditions. The developed model predicts energy consumption using vehicle speed and road-grade inputs for a drive cycle. The accuracy of the model is validated by comparing the prediction results against track and road test data. The developed model was able to accurately predict the energy consumption for track drive cycles within the error of ±4.0% of that measured from the experimental data. Finally, the model has been tested and verified for real-time implementation using the dSPACE MicroAutoBox II HIL test bench.