A sampling-based method for high-dimensional time-variant reliability analysis. (1st July 2019)

• Record Type: Journal Article
• Authors: Li, Hong-Shuang; Wang, Tao; Yuan, Jiao-Yang; Zhang, Hang
• Is Part Of: Mechanical systems and signal processing. Volume 126(2019)
• Page Start: 505
• Page End: 520
• Publication Date: 2019-07-01
• Subjects: Time-variant reliability analysis -- High dimensions -- Composite limit state -- Generalized subset simulation -- Series expansion methods -- Cumulative failure probability curve; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2019.02.050
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A sampling method is proposed for high-dimensional time-variant reliability analysis. Time-variant system is reinterpreted as multi-response time-invariant system. The proposed method is tested on two high-dimensional dynamic reliability problems. Results are compared to MCS and the original subset simulation. Abstract: A new sampling-based method is proposed for high-dimensional time-variant reliability analysis with both random variables and random process as inputs. The new method employs the series expansion methods, e.g., the Karhunen-Loève expansion, to represent the input random process into a set of random variables. Based on the concepts of composite limit state, the time-variant reliability analysis is converted into a series system reliability problem with multiple responses. Then the generalized subset simulation is applied to compute cumulative failure probabilities which are further used to interpolate a completely cumulative failure probability curve for a given time interval. The advantage of the proposed method is that only a single run can provide a cumulative failure probability curve instead of repeated runs. Two high-dimensional time-variant reliability problems with input random process are used to demonstrate the performance of the proposed method.