With the increasing number of phased antenna and the increasing size of airborne radar, the demand of rapid diagnosis in fault unit is becoming more and more urgent. Based on the principle of fault diagnosis for phased array antenna, a fault signal preprocessing method and a diagnosis method of phased array antenna is studied in this paper. At first, simulation environment is built with MoM. Secondly,using the method of the wavelet modulus maxima to achieve noise reduction processing and extracting the statistical features of time domain and wavelet energy features. Finally,using the improved SVM method to classify for getting a higher accuracy. As the scale increases, the classified accuracy does not decreased obviously. Results turn out that this method is effective in diagnosis of phased array element theoretically.