Accurate identification of cable tension has important practical significance, but the vibration-based tension estimation methods has the problem of complex and diverse calculation formulas in application. Based on this, this paper introduces the physics-informed neural network (PINN) into the field of cable tension identification, and proposes a method of cable tension identification based on physics-informed neural network. In this method, the loss term of string vibration equation representing the cable tension is introduced into the loss function of neural network, and the network is trained by using the dynamic displacement response of the cable. The cable tension can be identified by minimizing the loss function. Based on the finite test data of numerical simulation examples and indoor test model, the proposed method is used to realize the accurate identification of cable tension. The results show that the identification error of the proposed method is less than 3% in all the cable tension identification problems set in this paper. This result proves that this method has high identification accuracy, strong reliability and good scalability, and is suitable for the identification of cable tension in practical engineering.