To solve the problem that the convex hull coverage model of the dataset cannot reflect the effective distribution of the samples and the feature dimension of the sample points is too high in the process of fault diagnosis. A Sparse Scaled Convex Hull (SSCH) based support vector machine classification method is proposed and applied to fault diagnosis of roller bearings. The dimensionality reduction of the features of the sample set is carried out by random forest (RF). Firstly, the optimized sample subsets are obtained by sparse approximation, and the scale factor of the convex hull of the optimized sample set is adjusted so that the convex hulls of various sample sets are linearly separable. Secondly, to solve the problem of too high feature dimension, the importance of features is evaluated by random forest, and some redundant features are removed. Finally, the support vector machine model is constructed by the closest points between the convex hulls. The experimental results and related theories show that the proposed method has high classification accuracy in bearing fault diagnosis，the method can reliably identify different types of faults in rolling bearings.