In view of the insufficient exploitation of the massive data from the thermal power plant supervisory information system（SIS）， a similarity association rules data mining method based on principal component analysis（PCA） is proposed. Its applications in power plant parameter correlation analysis and abnormal data detection are discussed. Historical data from real-time database plant information（PI） of SIS through the .NET platform are analyzed by PCA in the MATLAB environment to find the similarity association measuring points. Data from a 600 MW unit show that the method proposed is capable of accurately and quickly seeking parameters with fairly high similarity association rules according to the operation of a specific unit， thus successfully utilizing these similarity association measuring points in different ways.