Methodology for global sensitivity analysis of consequence models

Gant, Simon; Kelsey, Adrian; McNally, Kevin; Witlox, Henk W.M.; Bilio, Mike

doi:10.1016/j.jlp.2013.02.009

Cited by 19 publications

(14 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Quantitative global sensitivity analysis is computationally expensive requiring many thousands of model evaluations. Computational cost and efficiency plays a crucial role in decisions regarding modeling approaches for global sensitivity analysis, particularly for complex systems with a large number of parameters and a long runtime for each simulation (McNally et al, 2011 ; Gant et al, 2013 ). An emulator based upon a Gaussian Process regression model was chosen over eFAST primarily due to computational efficiency: eFAST was not computationally feasible.…”

Section: Discussionmentioning

confidence: 99%

“…This involved the construction of a surrogate model, referred to in the literature as an emulator, and performing sensitivity analysis on the emulator. A workflow for the global sensitivity analysis of consequence models using an emulator has been described in Gant et al ( 2013 ); a similar process has been adopted in our work.…”

Section: Methodsmentioning

confidence: 99%

“…The efficiency of this technique arises from the use of an emulator, which is a statistical approximation to the BBDR model. The emulator used in this work was based upon a Gaussian Process regression model that has certain properties that are suitable for use with deterministic models (Oakley and O'Hagan, 2004 ; Kennedy, 2005 ; Gant et al, 2013 ). The emulator was built by using a relatively modest sample of model evaluations.…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Quantitative global sensitivity analysis of a biologically based dose-response pregnancy model for the thyroid endocrine system

et al. 2015

View full text Add to dashboard Cite

A deterministic biologically based dose-response model for the thyroidal system in a near-term pregnant woman and the fetus was recently developed to evaluate quantitatively thyroid hormone perturbations. The current work focuses on conducting a quantitative global sensitivity analysis on this complex model to identify and characterize the sources and contributions of uncertainties in the predicted model output. The workflow and methodologies suitable for computationally expensive models, such as the Morris screening method and Gaussian Emulation processes, were used for the implementation of the global sensitivity analysis. Sensitivity indices, such as main, total and interaction effects, were computed for a screened set of the total thyroidal system descriptive model input parameters. Furthermore, a narrower sub-set of the most influential parameters affecting the model output of maternal thyroid hormone levels were identified in addition to the characterization of their overall and pair-wise parameter interaction quotients. The characteristic trends of influence in model output for each of these individual model input parameters over their plausible ranges were elucidated using Gaussian Emulation processes. Through global sensitivity analysis we have gained a better understanding of the model behavior and performance beyond the domains of observation by the simultaneous variation in model inputs over their range of plausible uncertainties. The sensitivity analysis helped identify parameters that determine the driving mechanisms of the maternal and fetal iodide kinetics, thyroid function and their interactions, and contributed to an improved understanding of the system modeled. We have thus demonstrated the use and application of global sensitivity analysis for a biologically based dose-response model for sensitive life-stages such as pregnancy that provides richer information on the model and the thyroidal system modeled compared to local sensitivity analysis.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Quantitative global sensitivity analysis of a biologically based dose-response pregnancy model for the thyroid endocrine system

et al. 2015

View full text Add to dashboard Cite

show abstract

“…To deduce which of the MODIS-derived inputs produced a substantial effect on the target variable (GSR), two GEM-SA parameters were used: the main effect (ME) and the total effect (TE). The ME enumerates the influence of just one parameter varying in its own in relation to GSR, while the TE comprises the ME plus any variance due to possible interactions between that parameter and all of the other inputs varying at the same time [72]. Figure 2 is a "Lowry plot" and shows the relative contribution to the total variance in GSR, from each selected MODIS input.…”

Section: Data Preparation Feature Selection and Sensitivity Analysismentioning

confidence: 99%

Deep Learning Neural Networks Trained with MODIS Satellite-Derived Predictors for Long-Term Global Solar Radiation Prediction

et al. 2019

View full text Add to dashboard Cite

Solar energy predictive models designed to emulate the long-term (e.g., monthly) global solar radiation (GSR) trained with satellite-derived predictors can be employed as decision tenets in the exploration, installation and management of solar energy production systems in remote and inaccessible solar-powered sites. In spite of a plethora of models designed for GSR prediction, deep learning, representing a state-of-the-art intelligent tool, remains an attractive approach for renewable energy exploration, monitoring and forecasting. In this paper, algorithms based on deep belief networks and deep neural networks are designed to predict long-term GSR. Deep learning algorithms trained with publicly-accessible Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data are tested in Australia’s solar cities to predict the monthly GSR: single hidden layer and ensemble models. The monthly-scale MODIS-derived predictors (2003–2018) are adopted, with 15 diverse feature selection approaches including a Gaussian Emulation Machine for sensitivity analysis used to select optimal MODIS-predictor variables to simulate GSR against ground-truth values. Several statistical score metrics are adopted to comprehensively verify surface GSR simulations to ascertain the practicality of deep belief and deep neural networks. In the testing phase, deep learning models generate significantly lower absolute percentage bias (≤3%) and high Kling–Gupta efficiency (≥97.5%) values compared to the single hidden layer and ensemble model. This study ascertains that the optimal MODIS input variables employed in GSR prediction for solar energy applications can be relatively different for diverse sites, advocating a need for feature selection prior to the modelling of GSR. The proposed deep learning approach can be adopted to identify solar energy potential proactively in locations where it is impossible to install an environmental monitoring data acquisition instrument. Hence, MODIS and other related satellite-derived predictors can be incorporated for solar energy prediction as a strategy for long-term renewable energy exploration.

show abstract

“…Sensitivity analysis investigates the effect of perturbing input parameters on the output of a model (Tur anyi, 1997). Knowing this effect is especially important when parameters are poorly defined or feature a large range of variability (Gant et al, 2013). Performing sensitivity analysis is useful for any system for which the number of parameters involved, combined with the complex behaviour of the system, hinders the understanding of how certain parameters influence the results of interest.…”

Section: Parameter Transformation and Sensitivity Analysismentioning

confidence: 99%

Response surfaces for advanced consequence models: Two approaches

Braatz

Hisken

2017

Journal of Loss Prevention in the Process Industries

View full text Add to dashboard Cite

Validation of the surrogates at the overall optimal parameter values.. . C.4 Validation of the overall optimised model, separately for each campaign. C.5 Maximum overpressure across the rig for BFETS Phase 2 experiments for the standard and overall optimised model and experimental data.. . C.6 Overpressure-time histories for BFETS Phase 2 experiments for the standard model, overall optimised model and experimental data.. .. .

show abstract

Methodology for global sensitivity analysis of consequence models

Cited by 19 publications

References 19 publications

Quantitative global sensitivity analysis of a biologically based dose-response pregnancy model for the thyroid endocrine system

Quantitative global sensitivity analysis of a biologically based dose-response pregnancy model for the thyroid endocrine system

Deep Learning Neural Networks Trained with MODIS Satellite-Derived Predictors for Long-Term Global Solar Radiation Prediction

Response surfaces for advanced consequence models: Two approaches

Contact Info

Product

Resources

About