Optimal driving during electric vehicle acceleration using evolutionary algorithms

• Published in: Applied soft computing Vol. 34; pp. 217 - 235
• Main Authors: Chakraborty, Debasri, Vaz, Warren, Nandi, Arup Kr
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.09.2015
• Subjects: Acceleration duration; Acceleration energy; Driving comfort; Electric vehicle; Multi-objective optimization; Optimal driving; Acceleration energy; Optimal driving; Electric vehicle; Multi-objective optimization; Driving comfort; Acceleration duration
• ISSN: 1568-4946; 1872-9681
• DOI: 10.1016/j.asoc.2015.04.024

•A multi-objective based system proposed for comfortable driving strategy.•Energy consumption, acceleration duration, and jerk considered as objectives.•Crowding distance based NSGA-II was found appropriate to solve the problem.•Role of decision variables on system results analyzed using innovized principle.•The system found to be feasible for online implementation in electric vehicle. Due to the limited amount of stored battery energy it is necessary to optimally accelerate electric vehicles (EVs), especially in urban driving cycles. Moreover, a quick speed change is also important to minimize the trip time. Conversely, for comfortable driving, the jerk experienced during speed changing must be minimum. This study focuses on finding a comfortable driving strategy for EVs during speed changes by solving a multi-objective optimization problem (MOOP) with various conflicting objectives. Variants of two different competing evolutionary algorithms (EAs), NSGA-II (a non-dominated sorting multi-objective genetic algorithm) and SPEA 2 (strength Pareto evolutionary algorithm), are adopted to solve the problem. The design parameters include the acceleration value(s) with the associated duration(s) and the controller gains. The Pareto-optimal front is obtained by solving the corresponding MOOP. Suitable multi-criterion decision-making techniques are employed to select a preferred solution for practical implementation. After an extensive analysis of EA performance and keeping online implementation in mind, it was observed that NSGA-II with the crowding distance approach was the most suitable. A recently proposed innovization procedure was used to reveal salient properties associated with the obtained trade-off solutions. These solutions were analyzed to study the effectiveness of various parameters influencing comfortable driving.