XBand DualPolarized radar has great advantages in precipitation hydrometeor identification. At present, the commonly used method is the fuzzy logic algorithm, which has better recognition ability. This method mainly depends on the membership function of the hydrometeor but it needs to adjust the suitable membership function for the aging problem of different radar sets. To solve this problem, this paper proposes the method of cluster analysis for precipitation phasestate recognition. After the phasestate recognition, the distribution of different radar parameters can be counted as the suitable membership function for the radar. The cluster analysis method performs well in the hydrometeor identification experiment, but there are still some problems. Some hydrometeor cannot be identified without the introduction of temperature, which is described.