Research Of Multi-fault Diagnosis Method Of Rotating Machinery Based On EMD And ICA-R

The movement and power transmission of rotating machinery is completed bytransmission and rotating parts, they are widely used in petroleum industry, chemicalindustry, electric power industry, aerospace industry, mechanical processing, etc.Most of the rotating machinery contain gears, bearings, shafts and other built-upmembers, when these parts fail, it will lead to accidental shutdown and casualtiesaccident, thus research of process monitoring and real-time fault diagnosis for healthstatus of the equipment is extremely significant.The major components of a rotating machinery interact with each other, in mostcases the faults are concurrent, under the influence of the uncertainty of transmissionpath and environmental noise, the traditional time-frequency analysis methods can’tseparate or extract the fault signals. In order to perform multi-fault diagnosis ofrotating machinery, a method of multi-fault diagnosis of rotating machinery based onempirical mode decomposition (EMD) and independent component analysis withreference (ICA–R) is proposed, and the results of simulation experiments and gearbox failure experiments verified its effectiveness. The major contents of this thesisincludeWe have explored the mixed mode of rotating machinery faulty vibrationsemaphores, combine all these factors include the structure, dimensions and stiffnesscharacteristics of the experiment equipment, we established a linear instantaneousmixed fault simulation signal based on the response model of gear and bearing fault.A developed algorithm of EMD based on the characteristics of white noise toextract vibration mode is applied to preprocess the mechanical vibration signal,achieved the channel extension of single-channel mechanical signal mixture, faultfeature extraction by using ICA-R method. The result of simulation experimentsproved the availability of the proposed method.Exploited a GUI software used for rotating machinery mixed fault featureextraction depended on our study. For a collected gearbox multi-fault signal which issimulated by the gearbox dynamic simulation system, blended by the vibrationsignals of gear, bearing failure, the multi-fault diagnosis method of rotatingmachinery based on EMD and ICA-R is used to extract equipment failureinformation, compared with the outcome of FASTICA, it shows that the method isavailable to extract target vibration signal, realize failure diagnosis.