An Efficient and Time-Saving Reliability Analysis Strategy for Complex Mechanical Structure

Abstract

With the development of production technology, the mechanical structure has become more and more complicated, which makes the simulation process of the corresponding computer model very time-consuming. As a result, the reliability analysis needs to consume huge time resources. To deal with this problem, an improved method, which balances the accuracy of failure probability and the efficiency of computation, is proposed. The novelty of the proposed method is the use of an efficient point selection strategy, the k-means algorithm, and the constraint of correlation among the training points. The k-means algorithm can divide the candidate points into a few groups. Therefore, we can update the Kriging model by selecting several sample points which have large contributions to improve the accuracy of failure probability in each iteration. Meanwhile, a constraint is applied to control the relative location among points of DoE to avoid redundant information. The efficiency and accuracy of the proposed method are verified through two numerical examples. Finally, a type of artillery coordination system as a representative of the complex mechanisms is mentioned. And the proposed method is applied to calculate the reliability of the position accuracy of the coordination process, which proves the significance of the proposed method in engineering practice.

Importance Sampling for Time-Variant Reliability Analysis

“…Time-independent reliability analysis has been extensively studied. Various methods, including first and second order reliability methods (FORM & SORM) [2], random simulation methods [3,4], surrogate model-based methods [5][6][7], etc., have been developed during the past decades. FORM and SORM are quite efficient, but they may provide inaccurate results when performance functions are highly nonlinear.…”