一种预测随机谱下裂纹扩展曲线的新方法  

A New Method for Predicting Crack Propagation Curve under Random Load Spectrum

作　　者：潘绍振 刘小冬[1] 董江[1] 兑红娜 

机构地区：[1]成都飞机设计研究所,成都610091

出　　处：《机械科学与技术》2017年第7期1143-1148,共6页Mechanical Science and Technology for Aerospace Engineering

摘　　要：提出了一种基于贝叶斯理论在随机谱下采用Walker公式预测裂纹扩展曲线的新方法。首先,将Walker公式中裂纹扩展参数C、n视为随机变量,利用随机谱下的试验数据,基于贝叶斯理论建立其联合后验分布表达式;其次,巧妙地将随机谱下的裂纹扩展分析嵌入马尔科夫链蒙特卡洛(MCMC)方法中,实现对C、n后验分布的抽样;最后,将C、n后验分布均值代入Walker公式中,预测给定初始长度的裂纹在随机谱下的扩展曲线(a-N曲线)。利用7050-T7651和7050-T7452两种材料在随机谱下的裂纹扩展数据验证,发现仅需使用较少的试验数据,基于C、n后验均值预测的a-N曲线与试验a-N曲线就能良好吻合。研究结果对实现结构健康监控(SHM)中对结构未来损伤的准确预测,具有较大的工程应用价值。A Bayesian approach based crack propagation curve predicting method under random load spectrum with Walker model was presented. At first, parameters C and n in Walker model were viewed as random variables and their joint posterior distribution was constructed based on Bayes＇ theorem using test data under random spectrum. Then, samples of C, n were generated from their joint posterior distribution using Markov Chain Monte Carlo （MCMC） sampling method in which crack growth analysis under random spectrum was embedded ingeniously. At last, mean values of posterior distribution of C, n were substituted into Walker model to predict crack propagation curve （a-N curve） whose initial size has been known under random spectrum. The presented method was identified using crack growth test data of Aluminum alloy 7050-T7651 and 7050-T7452 under random spectrum, it was found that a-N curve predicted based on the method agrees well with tested a-N curve even only using a few of test data. Hence, it＇s promising that this method would bring great value in engineering application for achieving future damage predicting accurately in structural health monitoring （SHM） system.

关 键 词：Walker公式 贝叶斯理论 结构健康监控 马尔科夫链蒙特卡洛(MCMC)方法 裂纹扩展 

分 类 号：V215.5[航空宇航科学与技术—航空宇航推进理论与工程] O346.1[理学—固体力学]