Experimental Analysis and Model-Based Characterization of the Toyota Mirai II's Electric Machine

The electrification of the transportation sector relies on extensive research and data availability to accelerate technological advancements. However, for certain key components such as electric machines, detailed operational information remains scarce, which in turn limits the development of accurate system-level models for electrified powertrains. As a contribution to addressing this challenge, this study presents an experimental benchmark of the electric machine in the second-generation Toyota Mirai, a fuel cell hybrid electric vehicle (FCHEV) featuring a variable DC voltage bus, which was tested on a roller test bench. The proposed methodology aims to characterize the electric machine with minimal instrumentation and prior knowledge of the machine's configuration, by identifying electrical and geometric parameters that are relevant for a steady-state model of the machine, applicable to system-level studies, with the objective of providing a methodology that can be used in future research in order to expand the availability of electric machine data in the literature. The study provides a discussion on the necessary assumptions to derive the required electrical quantities for the characterization, besides examining how the Mirai exploits DC bus voltage variation coupled with a modulation strategy transitions from a conventional space-vector modulation approach to full overmodulation with six-step operation. The results of the fitting procedure to identify the machine's parameters indicate that the estimated electrical quantities closely align with experimental data, with errors remaining under ten percent. The study also showed that the fitted parameters and the model estimated the cumulated energy losses during the machine's operation with errors below 1.5\% across all validation cycles. Limitations on the methodology are also presented, with recommendations for future studies as to which aspects of the methodology should be improved in order to increase its reliability and applicability to more general case studies.