| Summary: | 碩士 === 國立臺灣科技大學 === 電機工程系 === 107 === Rapid advances in algorithms and computing power recently paved new ways of prognostics and health management. Machines are vital assets for most industries. Its early fault diagnostics not only provide economic benefits to companies but also in protecting lives. A novel motor fault diagnosis using only motor current signature is developed using frequency occurrence plot-based convolutional neural network (FOP-CNN). In this study, a healthy motor and four identical motors with synthetically-applied fault conditions – bearing axis deviation, stator coil inter-turn short circuit, rotor broken strip, and outer bearing ring damage are tested. A set of 50 three-second sampling stator current signals from each motor fault condition are taken under five loading variations – 0, 25, 50, 75, and 100% artificial coupled load. A total of 750 sampling signals are collected and processed into frequency occurrence plots (FOPs) which later serve as CNN inputs. This is done first by transforming the time-series signals into its frequency spectrums then convert these into two-dimensional FOPs. Five-times stratified sampling cross-validation is performed. When motor load variations are considered as input label, FOP-CNN predicts motor fault conditions with 92.37% classification accuracy. It precisely classifies and recalls bearing axis deviation fault and healthy conditions with 99.92% and 96.13% f-scores, respectively. When motor loading variations are not used as input label, FOP-CNN still satisfactorily predicts motor condition with 80.25% overall accuracy. FOP-CNN serve as a new feature extraction technique for time-series input signals such as vibration sensors, thermocouples and acoustics.
|
|---|
