Fundamental Evaluation of Data Clustering Approaches for Driving Cycle-Based Machine Design Optimization

Abstract

This article presents a fundamental evaluation of different clustering methods for analyzing driving cycles toward the efficient design of electric machines. It uses clusters of operating points to identify representative points (RPs) and the related optimization weights to design electric machines with optimal efficiency in the specific operating range. Typically, the design optimization of machines is carried out for one or a few speed–torque points. This article shows that for a predictable driving cycle or a specific set of operating points on the torque–speed plot, cluster analysis can be used to determine the representative operating points, which can be utilized in the optimization process to improve the overall efficiency for the considered driving cycle or operation profile. Furthermore, if multiphysics design is considered, the machine performance metrics can be guaranteed in multiple domains. This article presents a review of data clustering methods and their application in machine design optimization, where the pros and cons of the methods are weighed up. Further, X-Means method is proposed as an automated approach for cluster analysis and identification of RPs. In order to assess the effectiveness of the proposed method, a case study is carried out for the electromagnetic design of an interior permanent magnet (IPM) machine for the Worldwide harmonized Light vehicles Test Procedure (WLTP) driving cycle.