Deep Belief Network-based Prediction for Gear Noise

Considering that the vibration and noise data of the gearbox had fewer characteristic parameters, the octave analysis method was used to expand the dimension of the characteristics. The fully coupled model of the gearbox solved the noise of the gearbox, and the reliability of the octave analysis data was verified by means of experiment and simulation. Amplify acceleration data, load data, and noise data into 28-dimensional vibration and noise data by octave analysis. A DBN noise prediction model based on PSO was established, and multi-condition data was used for training and prediction. The results of this method were compared with the results of BP and SVM, this method shows better accuracy.

Automated summary

In this study, the octave analysis method was used to expand the dimension of vibration and noise data of the gearbox, and a DBN noise prediction model based on PSO was established. The reliability of the octave analysis data was verified through experiment and simulation, and the results showed better accuracy compared to BP and SVM methods.