A context-aware approach to construct driving cycles in a volatile urban environment

Driving cycles are essential for estimating vehicle emissions, as regional driving dynamics are reflected through them. While exogenous factors such as weather conditions, road features, and type of day impact these patterns, this diversity is often neglected by many studies, and limited driving cycles are usually selected. In this study, a multifaceted framework was generated, resulting in a mixed model through which the influence of road and weather factors on driving cycle parameters was examined. A microtrip-based technique and the Markov process were employed on a comprehensive driving dataset consisting of over 10,567,000 trips to construct driving cycles tailored to different conditions, differentiating between weekdays and weekends. Subsequently, to determine the optimal number of driving cycles in volatile conditions, a two-stage method was developed, relying first on a similarity indicator to measure the similarity between cycles, and then on information loss and variability metrics to assess the variability of the driving cycles. It was found that Montreal's driving conditions are best captured by 76 distinct cycles.