.png){kind=icon}
返回
The exact error prediction method for MIMO controlled tests☆
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ymssp.2024.111877.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Multiple-input multiple-output (MIMO) tests are used to replicate environmental conditions on dynamic systems in a lab setting. MIMO tests allow responses at multiple locations to be replicated far better than possible with normal single input tests. However, achieving a desired response at multiple response locations is affected by various factors. Specifying test tolerances based on multiple responses (vs. on just a single input) is challenging and of great interest. The dynamic relationships between the inputs and the outputs of the system, given by the frequency response functions (FRFs), may not be perfectly characterized. Given a desired output, the FRF matrix must be inverted to obtain an input. This process has inexact solutions because the FRF matrix is usually not square and is ill-conditioned. The realization of time histories from a frequency domain input is not perfect, and the spectral content of the actual input to the system does not match what was initially sought after. Previous work has established a framework for predicting error in MIMO tests, the approximate error prediction method (AEPM), but was limited in the types of error sources that could be accounted for and the tests it could be applied to. This paper improves that method by changing to a matrix-based formulation and accounting for inversion error, called the exact error prediction method (EEPM). The EEPM is applied to similar tests in the previous paper, specifically single-input single-output (SISO) and square MIMO tests, with significant improvements in error prediction over the AEPM. Additionally, a broader set of rectangular MIMO tests, where inversion is a large source of error, are conducted with similarly effective results.
Keyword:
Vibration
Testing
Multiple-Input Multiple-Output
Error
Dynamics
期刊
IF:
8.9
论文数:
1.3W
被引数:
6.6W
.png){kind=icon}
