A Simulation-Ready Framework for Electric-Vehicle Gearbox NVH

Integrating Manufacturing Variability, Transmission Error, Structural Acoustics, Machine Learning, and Psychoacoustics

Abstract

Electric-vehicle (EV) gearboxes operate in an acoustic environment in which gear whine, motor orders, inverter-related harmonics, and structure-borne responses are readily perceived. Existing research often treats manufacturing variation, transmission error, multibody simulation, housing radiation, condition monitoring, and psychoacoustic assessment as separate topics. This review develops a simulation-ready framework that connects these domains in a single source-path-response-perception chain. The framework maps gear and assembly deviations, including pitch error, run-out, flank waviness, tooth-to-tooth microgeometry variation, bearing preload, misalignment, and housing stiffness, to intermediate quantities such as kinematic and dynamic transmission error, gear-mesh orders, shaft-order sidebands, structural vibration, sound-pressure bands, diagnostic indicators, and psychoacoustic metrics. A structured narrative synthesis of 38 sources combines 21 external publications with 17 representative publications from the author's research program to clarify which physical inputs and validation evidence are required for different engineering claims. Four simulation-readiness levels are proposed: nominal, tolerance-based, measurement-based, and variability-aware. These levels distinguish early concept screening from tolerance robustness, as-built unit reconstruction, and production-scatter prediction. The review also introduces a machine-readable source-claim representation to support transparent reuse of the cited evidence. The resulting framework provides practical guidance for robust EV gearbox NVH modeling, data-driven screening, root-cause analysis, end-of-line interpretation, and perception-oriented validation.