| Summary: | 碩士 === 國立臺北科技大學 === 自動化科技研究所 === 97 === This thesis proposes a robust fault diagnosis system of rotating machine adapting machine learning technology. The kernel of this diagnosis system includes a structure genetic algorithm neural network (sGANN). First, the frequency characteristics from differential fault signals are obtained by order tracking and full spectrum. The characteristic are used to feed into the sGANN corresponding to specified faults to emphasize the phenomenon of each fault. Especially, the structure genetic algorithm is applied to get the optimal parameters of the above sGANNs. In the final step of proposed diagnosis system, the evaluated indexes from sGANN are synthesized by a reasoning engine to identify the faults in the rotor system. In the experiment, six common malfunctions of rotor system, unbalance, misalignment, bow, rub, whirl and whip, are generated from a rotor kit to verify the performance of this diagnosis system. The advantage of this diagnosis system is that the optimal sGANN parameter can be automatically obtained, the local optimal can be reduced and the diagnosis accuracy can be improved.
|
|---|
