The Research Of Fault Diagnosis Of Water Pump Based On Wavelet Transform And Lipschitz Exponent

Water pump is the key equipment in many industrial processes. When the water pump, especially the boiler feed pumps and cooling pump of nuclear power station is in faulty condition, it will result in huge economic losses and serious social impact. To improve the reliability of water pump, the faults of pump must be timly diagnosised. At present, there are many methods for fault diagnosis of water pump, including spectrum analysis, power spectrum estimation, wavelet transform, fault tree and so on. Due to complexity and diversity of water pump faults, we can only know whether there is a fault and distinguish several faults, it is difficult to detect the aurae of pump fault.Considering that the vibration signals often contain singular components, when early fault is occured. So we can do research on fault diagnosis from the aspect of singularity analysis. Wavelet Transform Modulus Maximum (WTMM) which is developing from wavelet transform is a kind of effective way to analyse signal singularity, the degree of signal singularity can be reflected via Lipschitz exponent. Fault features are extracted by analysis and processing of vibration signal through the relationship of WTMM and Lipschitz exponent. We can obtain model knowledgebase by means of training lots of fault features. Suspected fault sets are obtained by matching fault features with model knowledgebase established in advance, and fault diagnosis is accomplished through intersection operation of suspected fault sets.The quality of vibration signal acquisition system directly affect the the reliability of pump fault diagnosis system. The acquisition system of vibration signal which was designed in this paper based on MMA7260 and uPD78F0547, the vibration signal was acquired by signal acquisition system, then transfered to the PC via RS-232 interface. We do research on water pump fault diagnosis which exemplified by overheating winding, imbalance and pedestal looseness, etc. The experiment results show that the method is effective in detecting early faults, determining fault types and obtaining the information of fault degree, which provide an effective way for fault diagnosis. It has the guiding significance to other diagnosis of complex machines.