Particle breakage plays an important role in determination of the physical and mechanical properties of granular materials. Limited by the unrepeatability and high expense of direct laboratory particle breakage tests, numerical methods (e.g., Discrete Element Method (DEM)) are often used to investigate the mechanism of particle breakage and corresponding soil macro-mechanical responses. However, the computational efforts of these numerical methods can be prohibitively expensive subjected to the situations, where a larger number of disk or sphere elements are used to approximately represent the soil matrix, or an accurate and high resolution prediction, where selection of micro input parameters needs thousands of trials. To address this difficulty, this study combines the Hermit Stochastic Response Surface (HSRS) with DEM analysis to predict the single particle breakage responses. The DEM analysis of particle breakage is first performed using PFC3D software with a set of predefined empirical input parameters. With this preliminary analysis, a fourth-order HSRS is established to predict the responses of single particle breakage. It is found that the fourth-order HSRS can predict the breakage characteristics of DEM simulation reasonably well. With the aid of HSRS, the calibration of micro input parameters can be accomplished with relative ease. It shows that DEM analysis using the parameters calibrated from HSRS can characterize the particle breakage behavior more realistically compared with that based on the conventional empirical parameters.