Energy Consumption Uncertainty Model For Battery-Electric Buses in Transit

Abstract

This study develops a Deep Learning Neural Network (DLNN) model to predict the consumed energy (EC) of Battery Electric Buses (BEBs) based on bus, route, driver aggressiveness, and environmental parameters. An ADVISOR simulation tool is utilized to estimate EC for 10,800 operation scenarios resulted from a fractional-factorial design. The scenarios are used in a DLNN model with a goodness-of-fit of 0.993. The results show that road gradient sharply increases the EC, while driver aggressiveness parameters considerably affect the EC. The outcomes provide a substantial indication for the operation of BEBs transit networks concerning the consumed energy.