Lithium-ion batteries are being used as energy storage systems in recent electric and hybrid elec-tric vehicles coming to market. Current cycle-life estimation techniques show evidence of discre-pancy between laboratory results and real-world results. This work is aimed at characterizing actual battery usage in electrified transportation applica-tions. Factors such as temperature, State Of Charge, Depth Of Discharge, current and voltage have to be carefully considered for accurate cycle-life prediction within a given application. Five electri-fied vehicles have been studied (two electric bicycles, one light EV, one mainstream HEV and one Heavy-Duty trolleybus). These vehicles have been equipped with sensors and data-logger and then test-driven on open roads under real-world conditions. Large amounts of data have been stored and later processed through an innovative method for analysis of actual usage. This method relies on data mining based on K-means clustering and synthetic duty cycle generation based on Markov chain modeling. Resulting synthetic cycles exhibit features similar to those observed on the large original datasets. This enables accurate prediction of cycle-life through realistic ageing trials of Lithium-ion batteries. Several experimental results are presented in order to assess the fitness of this method.
