The Research And Application Of Rotating Machinery Early Fault Diagnosis

The research is based on project “Research of Early Fault Diagnosis of Low-speedand Heavy-load Machinery based on Sparse Feature Identification”, supported byNational Natural Science Foundation of China.Rotating machinery is key equipment in the modern industrial production, theresearch of early fault diagnosis technology of rotating machinery has importanttheoretical significance and extensive application value. Due to the early fault signal oflow-speed and heavy-load machinery is very weak, it’s difficult to extract the faultcharacteristic frequency from the low frequency vibration signal and the strongbackground noise.In order to solve the problem which early fault characteristic frequency oflow-speed and heavy-load rotating machinery is difficult to extract. The furtherresearch focuses on the mechanism of early fault, the mathematical model of the pulsesignal, algorithm and the design of fault diagnosis platform. The dissertation proposedan adaptive morphology gradient lifting method. The Hilbert transform, adaptivemorphological gradient filter, and adaptive morphology gradient lifting method areused to analyze time domain and frequency spectrum of vibration signal. Comparingwith the three algorithms in extracting the weak pulse signal from vibration signalswith strong background noise, the adaptive morphology gradient lifting method isproved efficient under the low signal-to-noise ratio. The Research results, based onfault diagnosis platform, have obtained the good effect on early fault identification.On the hardware design, signal conditioning circuit does adaptive envelopedemodulation using the adaptive resonance demodulation technology to vibration signal,retains the pulse signal. DSP and ARM9dual core processing unit realize functions ofsignal processing, collection and control, automatic identification classification storage,network communication, multi-tasking, fault feature extraction and matching, etc.Software design of diagnosis platform, adopts embedded Linux operating system,builds the cross compile environment, transplants Boot Loader, kernel and root filesystem, codes device driver, design and transplant GUI, has functions of vibration signalspectrum analysis, fault diagnosis and data storage. The research results have been applied to industrial field, and have beensuccessfully used for the enterprise to provide the technical support that realizepredicting maintenance management strategy of mechanical equipment, createremarkable economic and social benefits.