ABSTRACT
A one-dimensional convolutional neural network and random forest-based fault diagnosis method for rotating machinery gears is proposed in this chapter. This method combines four types of sensor signals into a one-dimensional input signal and extracts signal features through a one-dimensional convolutional neural network. The feature matrix output is classified by the random forest classifier. From the comparison of experiment results, when the combined signal is used as the input, the diagnostic outcome is better than that of using a single vibration signal as the input. In addition, the experiment results also show that the random forest can provide better classification results when used as the classifier compared with those of softmax classifier and support vector machine classifier. In summary, the diagnosis method proposed in this chapter can effectively diagnose the faults of rotating machinery gears.
