With the high proportion of new energy units integrated into the power grid, the low inertia characteristic of the power system become more and more significant, which seriously affects the stability of the system operation. In order to accurately estimate the inertia of new energy power grid in actual operation state, a new energy power system equivalent inertia estimation method based on weighted recursive least squares (WRLS) -auto-regressive moving average exogenous(ARMAX) system identification is proposed. Firstly, a general inertia analytical model of generator power-frequency response characteristics under different disturbance conditions is established with the generator as the object. Secondly, the ARMAX model is established by taking the generator grid-connected bus active power and frequency disturbances as inputs and outputs. Considering that the actual power grid operation process is jointly affected by both large and small perturbations, the actual measurement data is heteroscedasticity, and the to-be-identified parameters in the model are solved by WRLS. Then, the transfer function model containing the inertia response in the identification model is extracted, and the inertia time constant of the inertia source is calculated using the step response, and the equivalent inertia of the system is calculated. Finally, the accuracy and practicability of the proposed method are verified by Matlab/Simulink simulation examples.