For the singularity diagnosis of deformation monitoring data of dams， conventional theories have low efficiency. In view of this situation， a new singularity diagnosis method is developed based on the principal component analysis （PCA） and co-integration analysis （CA）. Firstly， the squared prediction error （SPE） statistics are constructed according to PCA. The identification criteria of singular components are established by applying the hypothesis test principle. Secondly， the CA model is established and the estimation method of singular component is proposed by comprehensively using the augmented Dickey-Fuller （ADF） test and stepwise regression method. Finally， the validity of the proposed method is verified after analyzing engineering examples. Application scenarios show that： PCA criteria， Laida criteria， Dixon criteria and t criteria can identify isolated singular components with relative errors of 3.81%， 7.61%， 7.61% and 5.08%， respectively. The CA model has the highest accuracy in estimating the speckled singular components， followed by the statistical model and the autoregressive model. The multiple correlation coefficients are 0.994 5， 0.871 5， and 0.743 2， respectively. Compared with conventional methods， performances of the proposed PCA-CA methodology are improved， which provides an effective way for the singularity diagnosis of in-situ deformation monitoring data.