In order to identify and locate the degree and type of fault in hydraulic pipeline, a method of multivariate variational mode decomposition(MVMD) and weighted fusion of instantaneous phase characteristics and dominant frequency amplitude is proposed. The pipeline strain is monitored by distributed fiber Bragg grating(FBG), and the effective fault characteristics are extracted from the phase and amplitude of sensor signals, and the simulation experiments are carried out. The simulation results show that the pipeline fault features can be identified and located effectively according to the weighted fusion feature vector, and the accuracy of fault degree recognition is more than 98%.