中文摘要：

为了改善柔性机构动态可靠性分析的效率和精度,基于支持向量机SVM（Support Vector Machine）回归理论,提出了一种柔性机构动态可靠性分析高效率高精度的SVM回归极值法SREM（SVM Regression Extremum Method）。首先,介绍了柔性机构可靠性分析的基本理论;其次,融合蒙特卡洛法MC（Monte Carlo）和SVM回归理论,建立了柔性机构动态响应极值的代理模型,并利用代理模型进行柔性机构可靠性分析。最后,利用SREM法对柔性机构实例进行了可靠性分析,并与MC和人工神经网络ANN（Artificial Neural Networks）的分析结果进行比较。结果显示,在小样本情况下,进行柔性机构动态可靠性分析时,SREM的计算效率和计算精度都比ANN高;SREM的计算效率比MC大大提高,计算精度与MC相当。验证了在柔性机构可靠性分析中SREM的高效率和高精度,并证明了SREM在柔性机构可靠性分析中的可行性和有效行性。

英文摘要：

In order to effectively improve the efficiency and accuracy of dynamic reliability analysis in the Flexible Mechanism（FM）,based on the Support Vector Machine（SVM）regression theory,SVM Regression Extremum Method（SREM）is proposed to achieve the reliability of dynamic response in FM.Firstly,the basic reliability theory in FM was introduced.Secondly,the combination of Monte Carlo Method（MC）and SVM regression theory are applied to FM,and the surrogate model of dynamic response extremum in FM is established.Through using the surrogate model,dynamic response reliability in FM can be effectively implemented.Finally,one example for FM is conducted dynamic reliability analysis by SREM,by comparison with MC and Artificial Neural Networks（ANN）.The results show that,in the case of a small amount of samples,SREM is of higher precision and higher efficiency than ANN in the analysis of FM dynamic reliability;SREM is greatly higher efficient than MC,and SREM has almost the same accuracy as MC.SREM is proved to be of high efficiency and high accuracy in FM dynamic reliability analysis,and the feasibility and effectiveness of SREM are verified in the analysis of FM dynamic reliability.