Naturalistic driving cycle synthesis by Markov chain of different orders

Abstract

This paper evaluates the performance of using Markov chain of different orders to synthesise real-world representative drive cycles from numerous naturalistic drive cycles. The representative drive cycles can be a valuable input into the design of powertrains, especially for plug-in hybrid electric vehicle (PHEV) and electric vehicle (EV). Their onboard cost-sensitive electric components, such as battery, require an appropriate sizing by understanding how people drive in naturalistic settings. Applying representative drive cycles instead of federal certification drive cycles provides flexibility of drive cycle length and ensures realistic cycle aggressiveness. Even though Markov chain has been widely used to synthesise representative drive cycles, the effects of different orders have not been systematically compared. Based on a publicly accessible portion of GPS-enhanced regional household travel survey, after statistical hypothesis tests, the results show that higher degree of representativeness can be achieved with a 3-order Markov chain compared to a 2-order Markov chain. These findings help to improve the accuracy of cycle synthesis for PHEV and EV analysis.