In view of the problem that the current model of the model reference adaptive asynchronous motor based on flux linkage was influenced by the rotor time constant and the voltage model contained the pure integral link, a model that could identify the rotational speed and the rotor time constant at the same time was established, which improved the accuracy of the current model. The integrator containing two neuron adaptive filters was used to replace the integrator of the traditional voltage model, which eliminated the influence of the DC drift and the initial integration conditions. The simulation results showed that the proposed model reference adaptive system was not affected by the change of the rotor time constant and the DC drift, and the dynamic performance of the system was good. Finally, experiments were carried out on the DSP motor control experimental platform to verify the effectiveness and high performance of this method.