Ensemble empirical mode decomposition（EEMD） is presented to alleviate the mode mixing problem occurring in EMD.Feature information of signal is revealed more accurately than with EMD

with helps of EEMD.According to unstable-state and non-linear characteristics of acoustic emission signals

an applied method for tool wear state identification based on EEMD is presented.The IMF components with no mode mixing can be obtained with EEMD.The sensitivity evaluation algorithm extracts sensitive IMF from all the IMF.The energy of the sensitive IMF is extracted as input of support vector machine（SVM） classifier

and the tool wear state is divided into three kinds of state:normal cutting

medium wear and severe wear.By comparing classification accurate rate of EEMD and applied EMD methods

the superiority of the proposed method based on EEMD is demonstrated in state recognition of tool wear.