中文摘要：

考虑多裂纹间的相互干扰，建立含多裂纹结构的随机扩展模型，基于裂纹扩展随机过程的微熵最大化原理，采用随机梯度下降法推导出裂纹扩展速率的迭代式，从而计算出多裂纹扩展至指定尺寸所需加载循环数，并证明其与自然梯度算法等同且不受学习率的影响。通过风力机叶片裂纹扩展试验，记录多裂纹和单裂纹扩展时的扩展曲线，发现多裂纹干扰对主裂纹扩展有抑制作用，并用显微镜观察裂纹的金相组织，分析多裂纹增加断裂能的主要原因是层间纤维的强介质作用。最后在损伤容限分析中比较工程估算值、试验值以及该文提出方法的预测值，结果表明。该文所提出的方法可提高裂纹扩展寿命预测精度，提供风力机叶片损伤容限分析方法。

英文摘要：

The random propagation model of multiple cracks was set up by considering the interaction between multiple cracks. The iterative formula of the crack growth rate was derived using stochastic gradient descent method based on the entropy maximization principle for random processing of crack propagation. The required periodic cycles could be calculated for crack propagation using this method. It was proved that the algorithm was equivalent to the natural gradient algorithm and there was no impact on the learning rate. The propagating curves of the multiple cracks and the single crack were drawn ~hrough the crack propagating test. It was found that multiple cracks interference could restrain the main crack growth and the increased fracture energy of the multiple cracks was mainly due to strong inter-layer fiber effect by using the microscope to observe the microstructure of the cracks. Finally, the engineering estimation values, the test values and the prediction values of the proposed method were compared in the damage tolerance analysis. The resuhs showed that the proposed method can improve the prediction accuracy of the crack propagation life and the damage tolerance performance analysis method is given for the wind turbine blade.