Constructing the real-world driving cycle for electric vehicle applications: A comparative study

Abstract

Construction of real-world driving cycle is of great significance for designing and assessing the energy management strategy of electric vehicles. In this study, five methods of driving cycle construction are investigated, namely, the random selection method, principal component analysis-based method, clustering analysis method, Markov chain-based method, and the optimization-based method. Urban driving conditions in Shenyang, China, are used as a case study to construct the driving cycles using the five methods, respectively. Based on the above efforts, an evaluation method of driving cycle effectiveness is proposed from the three perspectives, namely, accuracy, operability, and reproducibility. Characteristic parameters, speed–acceleration probability distribution, impact on energy control effect, dependence on data volume, and result repeatability are considered specifically. The assessing results of the common methods are proposed by making systematic comparisons. The results disclose the advantages and disadvantages of each method and obtain the ranking of five methods in three performance indexes. The presented results are expected to provide a theoretical basis and useful guidance for establishing real-world driving cycles in practical engineering applications.