A correlation mining method for power quality data based on the trapezoidal cloud model is proposed. The division of level concept section with trapezoidal cloud model is able to soften the section boundaries by integration of randomness and fuzziness. The formulas for calculating the support and confidence based on the trapezoidal cloud model are defined for use in mining association rules from the power quality database. The correlation analysis with Kulczynski measurement and imbalance ratio will filter out senseless association rules. Through the mining and analysis of power quality data correlation

the operation characteristics hidden in the power grid can be found

and valuable information provided for decision-making by management. An example has verified the practicability and validity of the method proposed.