Assessment of ROC curves for inspection of random fields

Schoefs, Franck; Clément, Alexandre; Nouy, Anthony

doi:10.1016/j.strusafe.2009.01.004

Cited by 55 publications

(68 citation statements)

References 24 publications

(25 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Sensor measurements are considered as 'perfect' according to the definition provided in [Schoefs et al, 2009]. …”

Section: Main Assumptions For the Stochastic Modelingmentioning

confidence: 99%

Optimal sensor placement for Spatial Variability Assessment of Structures

Schoefs¹,

Bastidas‐Arteaga²

2017

Proceedings of the Joint COST TU1402 - COST TU1406 - IABSE WC1 Workshop: The Value of Structural Health Monitoring for the Reli

View full text Add to dashboard Cite

Abstract. Structural reliability assessment is largely influenced by the spatial variability of material properties or defaults; however, there are still various challenges for their characterization and modeling. Structural Health Monitoring (SHM) could provide useful information in space and time for spatial variability characterization of material properties and mechanical solicitations; nevertheless, this challenge is arduous because of the large number of potential sensor positions of local disruptions/failures. This paper proposes a methodology to optimize the spatial distribution of embedded sensors used for spatial variability assessment of stationary random fields. The optimization criterion relies on the width of the confidence interval of statistics for the characteristics to identify. For sake of simplicity, the paper illustrates the method for one-dimensional problems. The proposed method is applied firstly to a numerical example were several hypothetical structural configurations that could be found in practice are studied. It is finally applied to two case studies (a reinforced concrete beam and a steel wharf) where water content and loss of steel thickness are respectively measured. The results show that the stationary property is useful to deduce the minimum quantity of sensors and their position for a given quality requirement. They also allow us to propose a criterion for defining if regular or non-regular spacing of sensors along the inspection zone is more appropriate depending on the component length and autocorrelation structure of the random field.

show abstract

“…• Sensor measurements are considered as 'perfect' according to the definition provided in [Schoefs et al, 2009]. …”

Section: Main Assumptions For the Stochastic Modelingmentioning

confidence: 99%

Optimal sensor placement for Spatial Variability Assessment of Structures

Schoefs¹,

Bastidas‐Arteaga²

2017

Proceedings of the Joint COST TU1402 - COST TU1406 - IABSE WC1 Workshop: The Value of Structural Health Monitoring for the Reli

View full text Add to dashboard Cite

show abstract

“…More recently, BN have been applied as a flexible and powerful alternative to these models [26]. The requirement (b), to which we refer as inherent system reliability, is quantified through POD and PFA using the concepts of signal detection theory [27], which has found applications in many field including medical testing [28,29] and non-destructive testing of technical systems [30,31]. In the context of alarm systems, one can define the POD and the PFA as…”

Section: Reliability Of Alarm Systems For Natural Hazardsmentioning

confidence: 99%

Reliability and effectiveness of early warning systems for natural hazards: Concept and application to debris flow warning

Sättele

Bründl

Straub

2015

Reliability Engineering & System Safety

View full text Add to dashboard Cite

a b s t r a c tEarly Warning Systems (EWS) are increasingly applied to mitigate the risks posed by natural hazards. To compare the effect of EWS with alternative risk reduction measures and to optimize their design and operation, their reliability and effectiveness must be quantified. In the present contribution, a framework approach to the evaluation of threshold-based EWS for natural hazards is presented. The system reliability is classically represented by the Probability of Detection (POD) and Probability of False Alarms (PFA). We demonstrate how the EWS effectiveness, which is a measure of risk reduction, can be formulated as a function of POD and PFA. To model the EWS and compute the reliability, we develop a framework based on Bayesian Networks, which is further extended to a decision graph, facilitating the optimization of the warning system. In a case study, the framework is applied to the assessment of an existing debris flow EWS. The application demonstrates the potential of the framework for identifying the important factors influencing the effectiveness of the EWS and determining optimal warning strategies and system configurations.

show abstract

“…from measurements of a physical parameter (frequency, capacity, etc.). Nevertheless, their results are highly influenced by environmental conditions reducing its accuracy [12,13]. The accuracy of inspection results is largely improved by using destructive inspection techniques.…”

Section: Description Of the Strategy Of Maintenancementioning

confidence: 99%

“…It is supposed herein that the noise is independent of the real chloride concentration because there are several sources of error influencing the results of the inspection. This assumption has been validated from a series of measurements on metallic harbor structures [12]. It is also assumed that the distribution of the noise and the repair concentration are constant for each inspection.…”

Section: Consideration Of Imperfect Inspectionsmentioning

confidence: 99%