Aimed at the identification and analysis of hydrodynamic parameters of offshore platform, a platform motion model based on physical mechanism neural network is established by combining the platform motion data, and the hydrodynamic parameters are inversed. The results show that the neural network method can accurately fit and invert the numerical results. The test results show that the added mass parameters obtained are accurate, and the values of first-order non potential flow damping components at different frequencies are obtained, which provides a new method for identification and analysis of hydrodynamic parameters.