This paper investigates the problem of rotor flux observation in asynchronous motor based on the Super-twisting sliding mode algorithm. According to the asynchronous motor current model, one designes the open-loop estimation model of rotor magnetic chain under the super-twisting algorithm. In order to improve the robustness of the model to rotor resistance and external perturbations, the uncertainty part of the motor parameters in the observer is estimated. Furthermore, the rotor chain obtained by the observer is introduced into the stator current estimation equation, a closed-loop structure is constructed to improve the accuracy and convergence speed of magnetic chain observations. By reconstructing the intermediate variables to be observed in the model, the problems of integral drift and initial value error are figured out on the traditional observer model for solving the rotor magnetic chain. For the coupling problem of magnetic chain and rotor speed, a sufficient condition for the stability of the system is given by dealing with the perturbation. Simulation and experimental results verify the feasibility and effectiveness of the algorithm in this paper.