Abstract: Aiming at the problem of the long diagnosis time and the poor accuracy caused by the information redundancy or irrelevant information in the fault feature indicator sets from time domain,frequency domain and time-frequency domain of the doubly-fed wind turbine bearing,a new method for feature selection and parameter optimization of fault diagnosis is proposed,which combined Genetic Algorithm（GA）with Elman neural network（ENN）. In order to reduce the redundancy and irrelevant information,is used for feature selection to select the optimal feature subset. According to the minimum recognition error and the minimum number of feature subsets,the fitness function of ENN is constructed. To identify the bearing fault,the GA optimizes the ENN weight and threshold parameters,and then identifies the fault. The example results shown the effectiveness and accuracy of the method for fault diagnosis.