The reliability and controllability of laser delamination paint removal depend on effective online monitoring technology. The laser induced breakdown spectroscopy (LIBS) technology can effectively monitor the laser paint removal process. In this paper,laser is used to remove the paint layer on the surface of aircraft carbon fiber composite (CFRP). Based on the high repetition frequency laser paint removal LIBS online monitoring platform,60 groups of two kinds of spectra of topcoat and primer excited by the paint removal process are collected online. The discrimination and prediction models based on principal component analysis (PCA) and partial least squares (PLS) were established respectively,and the classification discrimination of LIBS spectrum in laser delamination paint removal process was studied. The cumulative contribution rate of the first two principal components of PCA model reached 79.2%,and the cumulative contribution rate of the first two principal components of PLS DA model reached 85.5%. The correction standard deviation (RMSEE) of the PLS regression model was 0.142923,the root mean square error (RMSEcv) was 0.152053,and the prediction standard deviation (RMSEP) of the model was 0.142421. The prediction accuracy was 100% for 20 groups of mixed data sets of laser cleaning topcoat and primer. The results show that PLS discrimination model is better than PCA model in classification discrimination,and PLS prediction model has better prediction accuracy in real time evaluation and automatic classification of paint layers. This research can provide technical support for LIBS to monitor the laser paint removal process online and realize automatic and intelligent laser paint removal.