Probabilistic estimation method for frequency response function level of noise and vibration on operational transfer path analysis using Markov Chain Monte Carlo Method

• Published in: Kikai Gakkai ronbunshū = Transactions of the Japan Society of Mechanical Engineers Vol. 85; no. 879; p. 19-00252
• Main Authors: MUTO, Daisuke, SATO, Wataru, YOSHIZAWA, Takashi
• Format: Journal Article
• Language: Japanese; English
• Published: The Japan Society of Mechanical Engineers 2019
• Subjects: Bayesian estimation; FRF level; Markov Chain Monte Carlo; Operational transfer path analysis
• ISSN: 2187-9761; 2187-9761
• DOI: 10.1299/transjsme.19-00252

Operational transfer path analysis is important in considering the countermeasure structure of the operating actual product for noise and vibration problems. But in the conventional method based on the least square method has some problems such as that reliability of the obtained results cannot be discussed. Therefore, we studied a method of Bayesian estimation for the FRF level and produced a trial calculation program to estimate by the Markov Chain Monte Carlo method, then verified the accuracy with the 2-input 1-output model. As a result, it was confirmed that the FRF level can be estimated with the reliability given as the probability distribution, and the expected value can be estimated with an error of 1.1 dB with respect to the true value of the preset FRF level. Next, in order to confirm the superiority of the proposed method to the conventional least square method, 10 sample data for evaluation with controlled inter dependencies are prepared, then FRF level estimated by proposed method has compared with that calculated by conventional method. Then, it was confirmed that the relation between FRF level estimated by each calculation methods and the independency of the prepared sample data. As a result, it was confirmed that all value estimated by proposed method show more accurately than conventional method.