Abstract: This paper proposes a novel Physics-Informed Neural Networks algorithm for linear elastic static crack problems. In order to handle the discontinuity of the displacement fileds along the crack, we adopt independent multi-neural networks to solve all the components of the displacement and stress fileds by using the domain decompositions technique. Moreover, in order to capture the singularity of the stress fields at the crack tip, we use the crack tip asymptotic field functions to modify the outputs of the multi-neural networks, which can improve the accuracy remarkably. Numerical experiments verify the efficiency of our algorithm.