Limits of variance-based sensitivity analysis for non-identifiability testing in high dimensional dynamic models

Dobre, Simona; Bastogne, Thierry; Profeta, Christophe; Barberi‐Heyob, Muriel; Richard, Alain

doi:10.1016/j.automatica.2012.05.004

Cited by 14 publications

(19 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The observed effects are especially interesting in the context of identification of unknown parameters. Knowing that non-influential parameters are non-identifiable [41], one can easily confirm that some of the model parameters might be identifiable using only some of the sensors, and only in some periods of time. For example, any attempt to identify initial Euler angles θ 0 and ψ 0 using magnetometer measurements H x , H y , H z after a few seconds of flight would be futile.…”

Section: B Results Interpretation For the Reduced Parameter Vectormentioning

confidence: 99%

“…• Provide insight into identifiability of the model's parameters, i.e. taking advantage of the knowledge that non-influential parameters are also non-identifiable [41]. Therefore, GSA might be used to answer the question if the uncertainty in sensor signals can be apportioned to any of the uncertain parameters -if not, they would be deemed non-identifiable.…”

Section: Global Sensitivity Analysis Theorymentioning

confidence: 99%

See 1 more Smart Citation

Global Sensitivity Analysis for Modeling the Free-Flight Behavior of an Artillery Projectile

Machala

Collin²,

Gilson³

et al. 2020

AIAA Journal

Self Cite

View full text Add to dashboard Cite

A vehicle's in-flight behavior can be represented by the Newton-Euler equations of motion: usually, such a model has a nonlinear and continuous description based on ordinary differential equations. The model structure can be altered using analytic transformations-such as transformation into a quasi-LPV structure-and, when the model is supposed to reflect an existing physical system, some of its parameters might be uncertain. This paper's objective is twofold: firstly, to investigate if a quasi-LPV model does accurately replicate the physical behavior of a large-caliber spin-stabilized projectile in free flight; and secondly, to verify if the available embedded sensors data could be used to determine non-identifiability of uncertain model parameters. In order to investigate these prospects, a permutation-based global sensitivity analysis is employed. The proposed study has highlighted the non-identifiable parameters of the model, either during the whole flight or some periods of it. These results can help to enhance the preparation of free flight experiments at the open-range test site of the French-German Research Institute of Saint Louis. Nomenclature M = Mach number V = airspeed, m s −1 α, β, α t = angle of attack, angle of sideslip, total angle of attack, deg C X , C N α , C y pα = axial force, normal force slope, and Magnus force slope coefficients C lp , C mα , C mq , C npα = roll damping, pitch moment slope, pitch damping and Magnus moment slope coefficients

show abstract

Section: B Results Interpretation For the Reduced Parameter Vectormentioning

confidence: 99%

Section: Global Sensitivity Analysis Theorymentioning

confidence: 99%

Global Sensitivity Analysis for Modeling the Free-Flight Behavior of an Artillery Projectile

Machala

Collin²,

Gilson³

et al. 2020

AIAA Journal

Self Cite

View full text Add to dashboard Cite

show abstract

“…13 by van Schepdael et al and Chap. 15 by Mengoni et al Even though the link between sensitivity and identifiability seems obvious, they are not exactly antonymous [13]. Thus, additional parameter identifiability analyses are necessary for designing optimal experiments [38].…”

Section: Practical Identifiabilitymentioning

confidence: 96%

“…This is achieved via the addition of a fluorescent dye that reacts with the molecule of interest allows for an indirect measure of that molecules concentration. This is for example employed for intracellular calcium (Ca 2+ ), or for 13 C mass spectrometry and a stoichiometric model of the extended central metabolism [22]. The upper value represents the net flux (in direction of the arrow), the lower value (on blue background) the backward flux.…”

Section: Microscopymentioning

confidence: 99%

See 1 more Smart Citation

The Experimental Side of Parameter Estimation

Schliemann-Bullinger¹,

Fey²,

Bastogne³

et al. 2015

Uncertainty in Biology

Self Cite

View full text Add to dashboard Cite

The development of mathematical models is an iterative process. One of its key steps is the acquisition of experimental data. For developing quantitative and dynamical models, this data needs to contain time-courses as well as several perturbation experiments. This article gives an overview of and an introduction into various experimental techniques frequently used during the process of developing dynamical models of biological signalling networks. Among others, the article discusses western blotting, enzyme assays, flow cytometry, protein mass spectrometry, DNA microarrays and different fluorescent microscopy techniques and stresses the significance of single-cell versus population measurements.

show abstract

Global sampling to assess the value of diverse observations in conditioning a real‐world groundwater flow and transport model

Delsman

Winters

Vandenbohede

et al. 2016

Water Resources Research

View full text Add to dashboard Cite

The use of additional types of observational data has often been suggested to alleviate the ill-posedness inherent to parameter estimation of groundwater models and constrain model uncertainty. Disinformation in observational data caused by errors in either the observations or the chosen model structure may, however, confound the value of adding observational data in model conditioning. This paper uses the global generalized likelihood uncertainty estimation methodology to investigate the value of different observational data types (heads, fluxes, salinity, and temperature) in conditioning a groundwater flow and transport model of an extensively monitored field site in the Netherlands. We compared model conditioning using the real observations to a synthetic model experiment, to demonstrate the possible influence of disinformation in observational data in model conditioning. Results showed that the value of different conditioning targets was less evident when conditioning to real measurements than in a measurement error-only synthetic model experiment. While in the synthetic experiment, all conditioning targets clearly improved model outcomes, minor improvements or even worsening of model outcomes was observed for the real measurements. This result was caused by errors in both the model structure and the observations, resulting in disinformation in the observational data. The observed impact of disinformation in the observational data reiterates the necessity of thorough data validation and the need for accounting for both model structural and observational errors in model conditioning. It further suggests caution when translating results of synthetic modeling examples to real-world applications. Still, applying diverse conditioning data types was found to be essential to constrain uncertainty, especially in the transport of solutes in the model

show abstract

Limits of variance-based sensitivity analysis for non-identifiability testing in high dimensional dynamic models

Cited by 14 publications

References 29 publications

Global Sensitivity Analysis for Modeling the Free-Flight Behavior of an Artillery Projectile

Global Sensitivity Analysis for Modeling the Free-Flight Behavior of an Artillery Projectile

The Experimental Side of Parameter Estimation

Global sampling to assess the value of diverse observations in conditioning a real‐world groundwater flow and transport model

Contact Info

Product

Resources

About