A Bayesian method for load identification of powertrain mounting system with interval uncertainty

2022-01-0746

03/29/2022

Event
WCX SAE World Congress Experience
Authors Abstract
Content
It is very important to accurately identify the load of the vehicle in the working state for the study of vehicle vibration and noise. At present, the identification technology of dynamic load mainly focuses on the deterministic model, but there are always structural model errors in practical engineering. Bayesian method and interval theory can deal with unknown-but-bounded uncertainties in measurement noise and structural systems. In this paper, a time domain load identification method based on Gibbs sampling and a hybrid dynamic load identification method based on Bayesian and interval analysis are proposed. Firstly, the structural model error is considered as the overall Markov parameter matrix error, and the measurement noise is calculated together; secondly, considering the structural model error as interval parameter, the upper and lower load boundaries are obtained by combining interval analysis with Gibbs sampling method based on Taylor expansion. It is shown that the Gibbs sampling method based on the traditional state space equation is more sensitive to the type of input load under the action of three excitation forces at the same time, which means that the measurement of the fast changing pulse load requires less environmental noise and higher sampling frequency. The dynamic pair mass parameters are more sensitive than stiffness, and the uncertainty of dynamic load increases with the increase of parameter uncertainty. Therefore, this method is suitable for the case of less uncertainty. Finally, the impact test of force hammer verifies the universality of the proposed method. Keywords: Load identification; Gibbs sampling method; Bayes; Interval analysis
Meta TagsDetails
Citation
gao, b., and Liu PhD, X., "A Bayesian method for load identification of powertrain mounting system with interval uncertainty," SAE Technical Paper 2022-01-0746, 2022, .
Additional Details
Publisher
Published
Mar 29, 2022
Product Code
2022-01-0746
Content Type
Technical Paper
Language
English