In order to overcome the difficulty in Continuous Commutation Failure (CCF) early warning caused by the lack of CCF analytical mechanism, a CCF warning system is proposed based on Adaboost algorithm in statistical learning modeling. The eigenvector of the classifier considers the analytical expression of single commutation failure, and makes full use of the measured data during commutation failure from simulation or actual data from industrial field to obtain the nonlinear decision plane that corresponds to CCF early warning problem. The model verification is evaluated based on multi-scene fault simulation with CIGRE standard model in PSCAD, where the proposed warning system is able to send out warning signal with high precision based on the considerably small data set as well as sufficiently small measured data required, proving the validity and practicability of the proposed method in engineering practice. This work is supported by National Natural Science Foundation of China (No. 51477084) and Science and Technology Project of State Grid Shandong Electric Power Company “Application and Engineering Demonstration of Global Energy Interconnection based ‘Source-Grid-Load’ Coordinated Control Technology in Large-scale Safe and Stable Running” (No. 2017A-58).