DATA-DRIVEN ENERGY ECONOMY PREDICTIONS FOR- ELECTRICITY BUSES USING MACHINE LEARNING

Abstract

Electrification of transportation systems is increasing, in particular city buses raise 
enormous potential. Deep understanding of real-world driving data is essential for 
vehicle design and fleet operation. Various technological aspects must be considered 
to run alternative powertrains efficiently. Uncertainty about energy demand results in 
conservative design which implies inefficiency and high costs. Both, industry, and 
academia miss analytical solutions to solve this problem due to complexity and 
interrelation of parameters. Precise energy demand prediction enables significant cost 
reduction by optimized operations. This paper aims at increased transparency of 
battery electric buses’ (BEB) energy economy.We introduce novel sets of explanatory 
variables to characterize speed profiles, which we utilize in powerful machine 
learning methods. We develop and comprehensively assess 5 different algorithms 
regarding prediction accuracy, robustness, and overall applicability. Achieving a 
prediction accuracy of more than 94%, our models performed excellent in 
combination with the sophisticated selection of features. The presented methodology 
bears enormous potential for manufacturers, fleet operators and communities to 
transform mobility and thus pave the way for sustainable, public transportation.