The variation in load on the piezoelectric positioning platform significantly limits its accuracy during rapid positioning. To address this， a robust disturbance observer control strategy was proposed， treating load variation as an external disturbance. Firstly， based on the linearization of the MPI hysteresis inverse model， the linear dynamic characteristics of the piezoelectric positioning platform were identified according to experimental data. Next， the system identification error was attributed to model uncertainty， and a less conservative upper bound was established in the frequency domain. Then， considering the impact of model uncertainty and load disturbances on system stability and performance， a robust disturbance observer based on μ-synthesis was designed. Finally， the experimental results show that the proposed method achieves a disturbance rejection bandwidth of 114 Hz， which is 109% higher than that of the traditional DOB method and 86% higher than that of the H_∞DOB method. For tracking a 70 Hz sinusoidal trajectory， the proposed method achieves an average relative error change of 0.001 3 under both loaded and no-load conditions， which is a 91.1% reduction compared to the traditional DOB method and a 77.2% reduction compared to the H_∞DOB method， demonstrating the effectiveness and superiority of the proposed method.