空調機配管の伝達経路解析による騒音発生源の可視化（PLSに基づく実稼働TPAによる音響潜在モードの抽出）

抄録

This study examines Operational Transfer Path Analysis (OTPA) for vibration-acoustic dynamics in complex vehicle systems. OTPA simplifies data collection by avoiding structural transfer function pre-measurement, overcoming traditional method limitations on complex paths. It uses multivariate regression for transfer function estimation but faces noise and multicollinearity. Conventional PCR-based OTPA for vibration-acoustic coupling often misidentifies dominant sound contributors by prioritizing high-energy vibrations, potentially overlooking critical low-energy components or the low radiation efficiency of high-energy ones. To address this, a Partial Least Squares (PLS) regression-based OTPA is proposed. PLS extracts 'acoustic latent modes' by maximizing covariance between reference vibrations and response sound, better linking vibrations to acoustics. The PLS-OTPA is verified to extract key sound-constituting vibration modes and shows improved accuracy over PCR methods in analyzing an air-conditioning outdoor unit's piping. This approach significantly refines transfer path analysis accuracy, advancing vibration and acoustic analysis.