Structural elements, such as stiffened panels and lap joints, are basic components of aircraft structures. For aircraft structural design, designers select predesigned elements satisfying the design load requirement based on their load-carrying capabilities. Therefore, estimation of safety envelope of structural elements for load tolerances would be a good investment for design purpose. In this article, a method of estimating safety envelope is presented using probabilistic classification, which can estimate a specific level of failure probability under both aleatory and epistemic uncertainties. An important contribution of this article is that the calculation uncertainty is reflected in building a safety envelope using Gaussian process, and the effect of element test data on reducing the calculation uncertainty is incorporated by updating the Gaussian process model with the element test data. It is shown that even one element test can significantly reduce the calculation uncertainty due to lacking knowledge of actual physics, so that conservativeness in a safety envelope is significantly reduced. The proposed approach was demonstrated with a cantilever beam example, which represents a structural element. The example shows that calculation uncertainty provides about 93% conservativeness against the uncertainty due to a few element tests. It is shown that even a single element test can increase the load tolerance modeled with the safety envelope by 20%.
[1]
Raphael T. Haftka,et al.
Accounting for Future Redesign in the Optimization of an Integrated Thermal Protection System
,
2012
.
[2]
Carl E. Rasmussen,et al.
Gaussian processes for machine learning
,
2005,
Adaptive computation and machine learning.
[3]
Haoyu Wang,et al.
Safety Envelope for Load Tolerance and its Application to Fatigue Reliability Design
,
2006
.
[4]
R. Scheaffer,et al.
Mathematical Statistics with Applications.
,
1992
.
[5]
Simon Rogers,et al.
A First Course in Machine Learning
,
2011,
Chapman and Hall / CRC machine learning and pattern recognition series.
[6]
Kyung K. Choi,et al.
Reliability-based design optimization with confidence level under input model uncertainty due to limited test data
,
2011
.
[7]
Yoshua Bengio,et al.
Pattern Recognition and Neural Networks
,
1995
.
[8]
Byung Man Kwak,et al.
CONCEPT OF ALLOWABLE LOAD SET AND ITS APPLICATION FOR EVALUATION OF STRUCTURAL INTEGRITY*
,
2002
.
[9]
A P Dawid,et al.
Properties of diagnostic data distributions.
,
1976,
Biometrics.
[10]
Nam H. Kim,et al.
How coupon and element tests reduce conservativeness in element failure prediction
,
2014,
Reliab. Eng. Syst. Saf..
[11]
Raphael T. Haftka,et al.
Effects of Structural Tests on Aircraft Safety
,
2010
.