There are many parameters in the cloud microphysics scheme and the constraint relationship between the parameters is complex, so it has not been decided what parameters should be selected and how to carry out the parameter combination disturbance in the ensemble forecast. Based on WRFv4.2 model, this paper selects eight parameters of WSM6 scheme to conduct the ensemble prediction experiment of parameter disturbance for the heavy rainfall event of Meiyu front in 2020. Contrast and analyze the influence of different parameters and different disturbance range combinations on prediction, and try to jointly disturb the sensitive parameters of cloud microphysical scheme and boundary layer scheme. The results show that the combined disturbance of the 4 sensitive parameters of the cloud microphysics schemes has the best performance in the prediction test, which reduces the False Alarm Rate of torrential rain probability forecast, and improves the spread skills of humidity field, zonal wind, temperature field in low level and meridional wind in middle troposphere. The combined disturbance of the raindrop truncation parameter and the maximum of the cloud ice diameter presents a synergistic resonance effect, which can effectively improve the ensemble prediction effect. However, combined perturbations between graupel truncation parameters, graupel density and truncation parameters show that disturbance energy does not increase with the increase of disturbance parameters, and the antagonism between parameters will restrict the improvement of ensemble forecasting skills. The disturbance of three parameters in the MRF boundary layer scheme has significantly improved the ensemble prediction skill of low layer humidity field, and the combined disturbance of three parameters is better than that of a single disturbance; At the same time, joint perturbation of the cloud microphysical parameter perturbation scheme and boundary layer parameter perturbation scheme can achieve the best effect. It shows that the selection of parameter and the range of the disturbance can affect the effect of the ensemble forecast, a collaborative relationship between the parameters of the combination is easy to obtain a better forecast effect, while parameter combinations with antagonistic effect show negative skills. Multi parameter disturbance considering the dependency relationship between parameters and the range of disturbance is more conducive to the improvement of ensemble prediction effect, which provides a valuable reference for the selection of parameters in parameter disturbance in the future.