A Study on Fault Diagnosis based on Machine Learning for Rotating Machinery from Vibration and Sound Signals

초록

Rotating machinery requires a reliable diagnosis method that accurately predicts operating state, since these systems are frequently operated in safety-related. This equipment generates various signals such as temperature, vibration, noise, and sound signals with operating. Research on using these signals to diagnose the operating condition of rotating machines is being actively conducted in industry and academia. In this study, a fault diagnosis for a rotating machine from vibration and sound signals was performed by machine learning. As machine learning algorithms, several algorithms such as SVM (Support Vector Machine), k-NN (K Nearest Neighbor), and MLP (Multi-Layer Perceptron) have been used. The accuracy of each algorithm is evaluated and MLP presents a reasonable result.