A systematic review of multivariate uncertainty quantification for engineering systems

Multistep prediction of dynamic uncertainty under limited data

“…Another key forecasting approach is deep learning, which makes use of neural networks (NNs) to learn from existing data. Applications are covered in detail for RUL prediction by Lei et al [12] and for uncertainty forecasting by Grenyer et al [10]. NNs are composed of multiple layers, allowing them to learn complex nonlinear relationships.…”

“…They must therefore be flexible to consider all data properties necessary to achieve robust predictions. Flexible models can make better predictions, but all predictions involve assumptions that manifest uncertainty [10,29,37]. Naturally, the structure of NNs and training options applied have a significant impact on prediction accuracy and robustness for specific applications [12].…”

“…Engineering systems are growing in complexity, consisting of numerous equipment units interacting in a linear or non-linear manner, with a collective behaviour distinct from and not often determined by that of individual units [7][8][9][10]. Applied in the engineering domain, these are termed complex engineering systems (CES).…”

“…The level of uncertainty can change throughout the in-service life. This calls for adaptive and intelligent approaches to forecast uncertainty based on a combination of available data and expert opinion [10].…”

Uncertainty is more than a number or colour: Involving experts in uncertainty assessments of yield gaps

“…Several methodologies exist for combining quantitative and qualitative uncertainty analysis. A review of multivariate uncertainty quantification for engineered systems is given by Grenyer et al (2021), while a review of different methods for modelling uncertainties is given by Elsawah et al (2020).…”