Martin Claesson och Erik Norheim, Elektroteknik
Title: Energy Prediction of Electric Trucks' Auxiliaries
Students: Martin Claesson and Erik Norheim Erik (MPSYS)
Industrial supervisors: Victor Olsson, Volvo GTT
Supervisor/Examiner: Balázs Kulcsar
In today's society a large proportion
of transportation is carried out on land by fossil-fueled vehicles but
there is an increasing trend towards electrifying vehicles. A
fundamental disadvantage of Electric Vehicles (EV) is the limited range.
An often overlooked aspect of the energy consumption is the
auxiliaries, which especially for trucks can be of a substantial
proportion of the total energy consumed. This thesis investigates data
driven methods to predict the auxiliary energy of electric trucks'
auxiliaries using historical data from Volvo GTT.
and prediction was done on preprocessed data to ensure that the results
are derived from feasible values of the signals measured. The analysis
laid the groundwork of determining the quality of the data and which
methods that were applicable on the problem. Results indicate that the
energy consumption of auxiliaries are difficult to predict with the
inputs available and does not always follow a typical nor expected
pattern, despite a significant correlation with the ambient temperature
and time. Furthermore, preprocessing of data proved to be a fundamental
process in enabling accurate predictions.
Testing models of
different complexity and types, the thesis found significant
improvements of the energy prediction compared to algorithms found in
relevant research papers when applied on the data. Machine Learning (ML)
models performed well considering the complexity of the problem, the
available signals and large amount of data. Lastly, important future
work is presented that can further improve the prediction of auxiliaries
and thereby contribute to more accurate range estimations.
Martin Claesson and Erik Norheim
E2 Room 5430 Femman, Chalmers