Research On The Incipient Fault Diagnosis Based On Fractal Theory For A Centrifugal Pump

The centrifugal pump is one of the most widely used fluid machinery. Some of these installations are crucial for a larger system to work. Faults of pumps can cause the breakdown of a whole system, and lead to substantial economic losses. This means the diagnosis of the incipient faults should play a very important role in industry. In recent years, there have been many studies on the diagnosis of the incipient faults, and some new methods were proposed, including the methods based on the fractal theory.In this dissertation, a new method based on the fractal theory and wavelet analysis is suggested for the diagnosis of the incipient faults of a pump system. The achievements of the dissertation are listed as follows.1. The domestic and foreign literatures were thoroughly reviewed. The studying significance and the present situation of the rotating machine fault diagnosis were discussed. Using National Instruments virtual instrumentation, a centrifugal pump testing rig was established to measure the vibration acceleration signal. The difficulties for the diagnosis of the incipient faults of a pump system are two factors: (1) The complex nonlinear relationships between the symptom parameters and fault types. (2) The difficulty of extracting the symptom of the fault embedded in strong noise.2. To detect abrupt change points, a method based on the exponent dynamical cross-correlation factor and the mean square value was presented. Ideal time series was used to evaluate the performance of the proposed method. CWRU vibration analysis showed that, compared with the load changes, the fault defects can cause much greater effects on the dynamic characteristics.3. To detect the fault feature and diagnosis, a method based on fractal method and wavelet analysis was presented in this dissertation. In this method, the vibration signals are wavelet transformed into different frequency bands at first. And then, based on the theory of the generalized dimension, the fault features are extracted. The CWRU and pump vibration analysis showed that this method is effective and easy to implement.