The nonlinearity inherent in the voltage source inverter (VSI) causes a deviation between the desired reference voltage and the actual output voltage of the VSI, which is particularly evident during the low-speed operation stage of the motor. In this paper, a novel approach is presented for online estimation of VSI nonlinearity by manipulating the permanent magnet (PM) flux linkage of the variable flux memory machine (VFMM). First, a voltage estimation model based on PM flux linkage disturbance considering VSI nonlinearity is developed. Second, based on this, in order to mitigate the impact caused by the error in rotor position, this method employs the dc components of q-axis equation and the observed PM flux linkage of VFMM to estimate the value of VSI-distorted voltage. The estimated distortion voltage is then fed back to the reference voltage to compensate for the nonlinearity of the VSI, thereby improving the dynamic performance of the system. The simulation results validate the effectiveness and viability of the proposed novel estimation approach.