Abstract: The power transformer is a key equipment in the power system,its operating condition is directly related to the reliability and safety of power supply. In order to make full use of multiple condition monitoring information of the power transformer and realize the accurate perception,the on-line monitoring parameters are analyzed by statistics analysis,and the correlation between monitoring parameters is determined.This paper proposes a PLS-PCA method for transformer’s condition analysis based on the regression characteristics of multi-state parameters. The statistical relationship among them is established by partial least squares regression（PLS）under transformer’s normal condition,and then the transformer’s condition evaluation is obtained by principal component analysis（PCA） based on the deviation value between the regression prediction and the real,the transformer’s condition change detection ability is compared by the control chart. Through the analysis of the monitoring data of the normal operating condition and abnormal condition of the transformer,it is shown that the multi-state parametric regression analysis can transform the multivariate monitoring data of the transformer into a statistical index,which can be displayed in an intuitive and visual way. It is easy to find the abnormal changes of the mean vector and covariance vector of the state parameters in the transformer operating process, and can detect the abnormal running condition of the transformer in advance.