中文摘要：

提出机构动态强度可靠性分析的理论和方法，给出柔性机构动态强度可靠性分析的模型。将驱动加速度、驱动时间、摩擦和阻尼力（矩）等作为随机变量，应用蒙特卡罗方法，取得动态参数样本，再利用人工神经网络方法，根据抽取的样本对网络进行训练，统计网络输出得到动态应力分布，进而求出机构动态强度可靠度。通过空间站柔性展开机构实例计算动态强度可靠度，结果表明，该方法精度高，与单纯使用蒙特卡罗方法的结果相比，提高了计算精度，大大减少计算时间，可用于复杂柔性机构的动态强度可靠性分析。

英文摘要：

The aim of the research is to present theory and methodology of flexible mechanism dynamical strength reliability analysis. A general model of dynamical strength reliability was introduced for flexible mechanism analysis. Stochastic variables included driven accelerations, driven time, torques, frictions and damps were considered basically. In the first, Monte Carlo （MC） method was applied to generate stochastic variables and dynamical responds of mechanism. Secondly, the application of artificial neural network （ANN） was motivated by the approximate concepts inherent in reliability analysis and time consuming repeated analyses required for MC. Finally, statistical distribution of dynamical stress was yielded from the outputs of ANN. As an example, one expand mechanism model of space station was employed to test this method. Compared with MC method, the results proved that this method could be used to account for the complicated dynamical strength reliability analysis in higher fidelity at a reasonable computational cost.