It is a main issue to find valuable information from the power quality (PQ) data beause of its big volume, heterogeneity and low value density in the power quality monitoring system of the grid. An analysis system of the power quality analysis based on the data mining technologies is presented in this paper, consisting of the technologies of the data cleaning, data fusion, cluster analysis, correlation analysis, and etc.. The proposed analysis system is applied in the power quality data analysis of a certain city power quality monitoring system. The meaningful variation laws of the power quality indices are obtained, which can provide valuable reference to the grid planning, dispatch and operation.