Due to the fact that only one adaptive coefficient could be adjusted in traditional Landau method for inertia identification, it was difficult to obtain both good identification speed and identification accuracy, leading to the instability of the speed adaptive control system in the case of large measurement noise. To solve this problem, a Landau method for inertia identification considering measurement noise was proposed. By using the same lowpass filter for both the speed and the electromagnetic torque, the reference model was the same as the one without filter. On this basis, both the adaptive coefficient and the time constant of the filter were adjusted for the same identification speed while with better noise suppression, so as to improve the inertia identification accuracy and the stability of the speed adaptive control system. The correctness and effectiveness of the proposed method were verified by simulations and experiments.