Abstract: The isothermal compression tests of GH4706 superalloy were conducted. The obtained true stress-true strain curves show complex deformation mechanism. Therefore, it is very important to establish an accurate constitutive model to characterize the flow behavior of GH4706 superalloy. Arrhenius constitutive model was established for GH4706 superalloy,and then was optimized by neural network training method to establish BP-ANN intelligent constitutive model. The results show that after ANN optimization for the constitutive model, the prediction error is reduced, and the ability of evaluation and prediction is greatly improved. The evolution of flow stress of GH4706 superalloy can be predicted accurately.