To optimize the efficiency of the linear compressor, its operating frequency must be controlled equal to the system resonant frequency. The traditional resonant frequency tracking control algorithm relies on the steady state characteristics of the system, which suffers from slow convergence speed, low accuracy and slow system response. In order to solve these problems, a novel resonant frequency tracking control for linear compressor based on model reference adaptive system (MRAS) is proposed in this paper, and the parameter adaptive rate is derived by the Popov's hyperstability theory, so that the system resonant frequency can be directly calculated through the parameter adaptive rate. Furthermore, the traditional algorithm needs to calculate the piston stroke signal by integrating the back-EMF, which has the problem of integral drift. The algorithm proposed in this paper only needs the velocity signal, and the accuracy of the velocity calculation can be ensured by utilizing the self-adaptive band-pass filter (SABPF), thereby greatly improving the accuracy of the resonance frequency calculation. Simulation results verify the effectiveness of the proposed algorithm.