A distribution density-based methodology for driving data cluster analysis: A case study for an extended-range electric city bus

Abstract

As each energy source in a multi-energy-source vehicle has its own high-efficient range, the fuel-saving performance of the vehicle is sensitive to driving conditions. In order to improve the vehicle fuel economy, the relationship of fuel consumption and driving condition type is needed to be established clearly as the basis of adaptive energy allocation strategy. This work proposes a two-steps methodology to establish such relationship through clustering micro-trips (which are segmented from driving data and used to indicate driving conditions) with a similar fuel consumption level together. The first step is to construct a fuel consumption-related feature vector by implementing correlation analysis for feature selection and applying principal component analysis for feature dimension reduction. The second step is to cluster micro-trips based on their density centers. This methodology is validated based on the driving data in Beijing, and three types of micro-trips are obtained from these driving data, and their average fuel consumptions are 8.5068, 12.7663 and 17.0686 L/100 km, respectively.