Taking the SCARA robot as the research object, the SCARA robot model was established in ADAMS software, and the simulation was carried out.The acceleration data of the crack prone parts of the SCARA robot were collected, such as the front and rear end of the big arm, the front and rear end of the small arm and the base; the fault diagnosis model of SCARA robot was established by using BP neural network in MATLAB to realize the intelligent recognition and classification for SCARA robot fault by using BP neural network. The results show that the calculation results of BP neural network are basically consistent with the expected output, and the accuracy and reliability are verified.