At present, multi parameter radar signal clustering is widely used, however, when the radar signals are heavily overlapped, there is a problem of low accuracy. Based on support vector clustering and delaminating coupling sorting algorithm, the standardized radar signal data are weighted by the variable precision rough set; by analyzing the results of cluster sorting, the model of effective evaluation is established; and the optimal clustering sorting parameters are determined. The simulation results show that when the radar signal data is seriously overlapped, the accuracy of the proposed method is improved greatly compared with the original method, which proves the effectiveness of the proposed method.