Research On Compound Fault Diagnosis Method Of Gear Box

The parts in the gearbox are not immediately overhauled as soon as they break down.The parts will be repaired or replaced when breakdowns reach a certain level of damage.In the process,the failure of one part usually causes the failure of other contact parts,and these failures form a compound fault.The compound fault is ubiquitous,and it is important to study the method of compound fault diagnosis to ensure the normal operation of the equipment.The gear and bearing are the two parts of the highest failure probability in the gear box.Therefore,this paper focuses on the study of the complex fault of the gear and bearing.When a gear box and a rolling bearing are at fault at the same time,the composite fault vibration signals often appear as nonlinear and unstable signals with multi-component coupling modulation.Combined with the development of gearbox compound fault vibration signal characteristics,current signal processing technology,feature extraction technology and pattern recognition technology,the paper adopts the following methods to diagnose composite faults.Firstly,the composite fault signal is decomposed by the LMD method based on the cubic segmentation Hermite interpolation,and the multi-component coupled modulation signal is decomposed into single component(PF component)modulation signal.Then,the envelope signal of single component modulation is analyzed by bispectrum,and the fault characteristic frequency of gear and bearing is extracted respectively.Finally,the SOM-BP complex neural network is established to identify the fault types of gear and bearing,and the energy value of different frequency segments is used as the feature vector.The experimental results show that the LMD method of three-time Hermite interpolation improves the fitting accuracy of the envelope.The validity of the diagnosis method of composite fault feature information is verified by the combination of rational Hermite interpolation of LMD and slice bispectrum.The accuracy of the fault type of gears and bearings is effectively recognized by the SOM-BP neural network model.