A novel uncertainty propagation and probability assessment method for the frequency response function involving correlated uncertainties

Baopeng Liao

DOI: https://doi.org/10.1007/s00419-024-02596-4

2024-04-27

Archive of Applied Mechanics

Abstract:Uncertainty propagation of the frequency response function is crucial for vibration problems such as model calibration, and the probability assessment is another indispensable item in structural optimization. Considering only one of them will inevitably lead to a lack of uncertain knowledge of practical engineering structures. However, it is still challenging to evaluate the system response bounds and probability characteristics simultaneously and maximally reduce the computational cost. This paper focuses on the frequency response function involving correlated uncertainties and proposes a novel uncertainty propagation and probability assessment method. First, a convex model was established to quantify the correlated uncertainty parameters, and then, the Chebyshev polynomial function was developed as the surrogate model to efficiently quantify the uncertainty propagation from uncertainty parameters to system response. Subsequently, the novel normalized and coordinated transformation combined the uncertainty propagation method, making the uncertainty system response easy to assess. Note that the response value at the interpolation point can be employed as the input of probability assessment. One can also estimate the probability characteristics at different frequency positions during the construction of the surrogate model. Finally, two numerical examples were presented to demonstrate the effectiveness and cheaper computation by a discrete system and a continuous system, respectively. Results indicate that the proposed method can be conveniently and accurately applied to assess the bounds and probability characteristics of frequency response function involving correlated uncertainties.

mechanics