Fault Diagnosis of Gear with Stress Wave Signal Based on EMD_CDBN

Accurate faults diagnosis of gear is great significance for avoiding serious mechanical accidents. The stress wave signals of gears contain abundant fault information. Therefore, this paper proposes the method for fault diagnosis about gear stress wave signals based on the EMD_ CDBN model, which can realize fault diagnosis and classification based on gear stress wave signals. Firstly, aiming at the problem of multiple coupled interference signals in gear stress wave signals, Empirical Modal Decomposition (EMD) is proposed to decompose the signal into multiple Intrinsic Mode Functions (IMFs). Then, based on Normalized Mutual Information (NMI) algorithm, the feature selection framework is constructed to select multiple IMFs based on the correlation. To select the best IMF signal, reduce data dimension, and improve model computation efficiency. Finally, fully combining the strong feature extraction capabilities of Convolutional Deep Belief Network (CDBN). The deep learning model is built to realize fault diagnosis of gear stress wave data. Research and experimental results show that the proposed method has good diagnostic results and high recognition accuracy for gear stress wave signals, with strong stability and generalization ability.