关键词: 权函数法, 应力强度因子, 深度学习, 偏心孔边穿透裂纹, 格林函数

Abstract: A deep-learning based multiple reference states weight function method is proposed to avoid the problems existing in tradi-tional multiple reference states weight function method for complex geometric configurations. Compared with the Green's functions at corresponding points of crack surface, the deviation between the Green’s functions of unilateral penetrating crack at the edge of the finite width plate obtained by this method and the Wu-Carlsson analytical weight function method is within 0.4%, which verifies the accuracy of the weight function obtained by this method and proves the feasibility of this method. Using this method, the weight function with some geometric dimensions is derived for edge crack at off-center hole in finite plate which is common in the aerospace field, and compared with the Green’s function value obtained by the finite element method, most of the deviations are less than 1.5%, and the maximum deviation is 4.8%, which further verifies the accuracy of the weight function solution obtained by this method. This method has the potential to be applied to complex geometric crack bodies.

Key words: weight function method, stress intensity factor, deep learning, crack at edge of off-center hole, Green’s function