Evaluating the use of rate-based monitoring for improved fatigue remnant life predictions

Leung, Michael Siu Hey; Corcoran, Joseph; Cawley, P.; Todd, Michael D.

doi:10.1016/j.ijfatigue.2018.11.012

Cited by 10 publications

(15 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…this same model will be used in this study to estimate probability density functions for t f for the "classical" linear FFM approach, in comparison to the Bayesian model that will be proposed and implemented in this work. More recent work has shown that the linearity assumption made in the regression implementation of FFM has been observed to be false for both early and late stage crack growth [24]. This work shows that imposing linearity may cause a positive bias in t f estimation for late stage crack growth, where the inverse feature rate has been observed to slope downward; this implies that the failure time occurs consistently sooner than predicted, introducing a non-conservative bias that could be catastrophic.…”

Section: Introductionmentioning

confidence: 76%

“…We will review the uncertainty model recently developed for this linear regression process presented in detail in [23,24]. Any given regression on some time-stamped feature set represents a single "block" observation, which is presumed representative of an ensemble population of regressions over the same time frame.…”

Section: The Linearized Ffm With Linear Regression Methodsmentioning

confidence: 99%

“…However, Refs. [23] and [24] used Eq. 7and showed it was sufficient to describe the distribution of the failure time, and it will be used here.…”

Section: The Linearized Ffm With Linear Regression Methodsmentioning

confidence: 99%

“…Table (1) shows a summary of the priors selected for the parameters. [22,24], inference is performed in the logarithmic space,…”

Section: Bayesian Modelmentioning

confidence: 99%

“…Bevilacqua's method is applied to earthquake eruption prediction, with small sets of data compared to the data presented in this study. In the domain of fatigue prediction, Leung et al [24] compared the performance between conventional periodic inspections informing Paris' law parameters (themselves updated via Bayesian inference) and continuous monitoring for use by the linearized formulation of the FFM. In that study, random uncertainty in the damage accumulation rate measurements is the only source of uncertainty for the FFM, which in turn results in uncertainties in the regression fit and the extrapolated time axis intercept (t f ).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A probabilistic estimation approach for the failure forecast method using Bayesian inference

O’Dowd

Madarshahian

Leung

et al. 2021

International Journal of Fatigue

Self Cite

View full text Add to dashboard Cite

Positive-feedback mechanisms such as fatigue induce a self-accelerating behavior, captured by models displaying infinite limit-state asymptotics, collectively known as the failure forecast method (FFM). This paper presents a Bayesian model parameter estimation approach to the fully nonlinear FFM implementation and compares the results to the classic linear regression formulation, including a regression uncertainty model. This process is demonstrated in a cyclic loading fatigue crack propagation application, both on a synthetic data set and on a full fatigue experiment. A novel "switch point" parameter is included in the Bayesian formulation to account for nonstationary changes in the growth parameter.

show abstract

Section: Introductionmentioning

confidence: 76%

Section: The Linearized Ffm With Linear Regression Methodsmentioning

confidence: 99%

“…However, Refs. [23] and [24] used Eq. 7and showed it was sufficient to describe the distribution of the failure time, and it will be used here.…”

Section: The Linearized Ffm With Linear Regression Methodsmentioning

confidence: 99%

“…Table (1) shows a summary of the priors selected for the parameters. [22,24], inference is performed in the logarithmic space,…”

Section: Bayesian Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A probabilistic estimation approach for the failure forecast method using Bayesian inference

O’Dowd

Madarshahian

Leung

et al. 2021

International Journal of Fatigue

Self Cite

View full text Add to dashboard Cite

show abstract

High-Rate Structural Health Monitoring and Prognostics: An Overview

Dodson

Downey

Laflamme

et al. 2021

Data Science in Engineering, Volume 9

Self Cite

View full text Add to dashboard Cite

Structural Health Monitoring (SHM) includes both static and highly dynamic engineering systems. With the advent of realtime sensing, edge-computing, and high-bandwidth computer memory, there is an ability to enable high-rate SHM (HR-SHM). The paper defines the technical area of high-rate structural health monitoring and prognostics and presents the HR-SHM technical grand challenges including: multi timescales of the problem, adequate sensor network and response, real-time assessment, and decision-making with quantified uncertainty and risk. Key issues to address in such challenges include the time duration of the event, time scales of the physics, multiple sources of uncertainty, as well as limited spatiotemporal constraints for hardware execution. The paper defines the high-rate time scale as 1 ms on the integrated paradigm including data acquisition, assessment execution, and decision-making. The spatial issues include the resolution of the area monitored, the communication distance, and the number of edge sensors. The temporal issue includes the sensor type (e.g., THz) as well as multiple sources of uncertainty. These constraints must be coupled to allow for high-rate implementation that is robust, adaptable, and beneficial to the missions of interest. To address the grand challenge, we propose physics-informed real-time fusion (PIRF) of high-speed dynamic data. Technologies such as machine learning and edge-computing can be further harnessed to enable structural and functional prognostics for high-rate dynamic systems. Quantification of uncertainty, both aleatory and epistemic, is necessary for real-time state estimation to be connected with the confidences to integrate risks into the decision-making.

show abstract

Diagnosis, Prognosis, and Maintenance Decision Making for Civil Infrastructure: Bayesian Data Analytics and Machine Learning

Vega

Yang

et al. 2021

Structural Integrity

Self Cite

View full text Add to dashboard Cite

Due to the aging of civil infrastructure and the associated economic impact, there is an increasing need to continuously monitor structural and nonstructural components for system life cycle management, including maintenance prioritization. For complex infrastructure, this monitoring process involves different types of data sources collected at different time scales and resolutions, including but not limited to abstracted rating data from human inspections, historical failure record data, uncertain cost data, high-fidelity physics-based simulation data, and online high-resolution structural health monitoring (SHM) data. The heterogeneity of the data sources poses challenges to implementing a diagnostic/prognostic framework for decision-making for life cycle actions such as maintenance. Using quoin blocks components of a miter gate as an example, this chapter presents a holistic Bayesian data analytics and machine learning (ML) framework to demonstrate how to integrate various data sources using Bayesian and ML methods for effective SHM, and Prognostics and Health Management (PHM). In particular, this chapter discusses how Bayesian data analytics and ML methods can be applied to (1) diagnosis of bearing loss-of-contact degradation in quoin blocks; (2) optimized sensor placement for SHM on the gate; (3) fusion of various data sources for effective PHM; and (4) deciding maintenance-strategies by considering the behavioral aspect of human decision-making under uncertainty.

show abstract

Evaluating the use of rate-based monitoring for improved fatigue remnant life predictions

Cited by 10 publications

References 21 publications

A probabilistic estimation approach for the failure forecast method using Bayesian inference

A probabilistic estimation approach for the failure forecast method using Bayesian inference

High-Rate Structural Health Monitoring and Prognostics: An Overview

Diagnosis, Prognosis, and Maintenance Decision Making for Civil Infrastructure: Bayesian Data Analytics and Machine Learning

Contact Info

Product

Resources

About