Vocation Identification for Heavy-duty Vehicles: A Tournament Bracket Approach

Abstract

The identification of the vocation of an unknown heavy-duty vehicle is valuable to parts' manufacturers. This study proposes a methodology for vocation identification that is based on clustering techniques. Two clustering algorithms are considered: K-Means and Expectation Maximization. These algorithms are used to first construct the operating profile of each vocation from a set of vehicles with known vocations. The vocation of an unknown vehicle is then determined by using one-versus-all or one-versus-one assignment. The one-versus-one assignment is more desirable because it scales with an increasing number of vocations and requires less data to be collected from the unknown vehicles. These characteristics are important to parts' manufacturers since their parts may be installed in different vocations. Specifically, this paper compares the one-versus-one bracket and the one-versus-one round-robin tournament assignments to the one-versus-all assignment. The tournament assignments are able to scale with an increasing number of vocations. However, the bracket assignment also benefits from a linear time complexity. The results show that despite its scalability and computational efficiency, the bracket vocation identification model has a high accuracy and a comparable precision and recall. The NREL Fleet DNA drive cycle dataset is used to demonstrate these findings.