This paper proposed a new method of time-domain signal feature extraction and fault diagnostic based on improved dynamic time warping method of mechanical and electrical equipment. Dynamic time warping method was improved by integrating the methods of phase compensation, weighted gradient, derivatives, sliding window connection, fast dynamic time planning . Then improved dynamic time warping method may be applied directly to extract the signals of domain statistical parameters of fault feature of mechanical and electrical equipments and get accurate residual signal sequence with fault sideband features. Identification and localization of fault signal features may be done by selecting peak value, RMS, kurtosis and other statistical characteristics identify and detect faults. Using new statistical parameter Thikat to forecast failure trend is a new method of fault detection and diagnosis of electromechanical equipments.