Research On Fault Diagnosis System Of Ship Generator Rotor

The widely used of the ship generator is the core part of the ship power system. It shows that theresearch in intelligent failure diagnosis of ship generator is very significant and can achieve the practicalvalue.According to the development of the generator fault diagnosis in the world, we make the deepresearch in the dissertation in generator common failure: bearing failure. In the online failure diagnosissystem of the ship generator bearing, the technology of wavelet neural network has been introduced andthe stator current d method has been used.Firstly, in the dissertation the common generator fault has been analysed, we do the main research inayalysing the reason of ship generator bearing fault and the characteristic frequency of the stator currentin fault. In the paper, the fault model of the generator stator has been build, the MATLAB analysismethod has been used to simulate the failures. It shows the inner connectivity of the characteristicfrequency of the generator stator current when the generator stator fault happened.Secondly, take the advantage of signal processing of the wavelet analysis, a new wavelet frequencyenergy method has been used to extract the eigenvalue of each frequency. The different points of thegenerator stator failure lead to the different energy eigenvalue. Hence, we can regard the each eigenvalueas the eigenvector in the failure diagnosis.Thirdly, according to the successful application in the failure diagnosis, we use BP neural networkmethod and Elman neural network method to diagnose the generator bearing fault. To compare the twomethods, we confirm to use the Elman network to diagnose the failure.Finally, the fault diagnosis model of the ship generator has been build. We used the neural networkto train the eigenvator which is extracted from wavelet analysis to construct the wavelet neural networkto recognize and diagnose the fault. The simulation of stator failure of shipboard generator bas been madeby the wavelet neural network system to complete the failure diagnosis.