In order to improve the accuracy of electrocardiogram（ ECG） classification

a ECG feature extraction strategy combining ensemble emprirical mode decomposition（ EEMD） and multiscale fuzzy entropy is proposed.ECG signal is decomposed into a series of IMF components by EEMD. Select effective IMF components to calculate fuzzy entropy of kinetic parameter and constitute the feature vector. Feature vector is sent to radial basis function（ RBF） neural network for training and recognization. Using data from MIT-BIH Arrthythmia Database for simulation. Experimental results show that the proposed algorithm is superior to single fuzzy entropy and EMD multi-scale fuzzy entropy method

and the generalization ability of the algorithm is superior to other algorithms.