Intelligent Diagnostic Study Using Machine Learning of Mechanical Faults in Rotating Machinery

In industrial maintenance, artificial neural networks are used to solve diagnostic problems by automatically classifying vibration signals corresponding to the different operating modes of machines. This study discusses the automatic diagnosis of mechanical defects of rotating machines by the Machine Learning method, based on the analysis of vibration signals from a test bench designed to study bearing defects. We also used machine learning as a means of classification and automatic diagnosis. Decision making on the mode of operation (without fault or with fault) of the system studied is done automatically by the decision tree method. The shape vector contains coefficients extracted from the signals taken from the test benches studied. The output vector contains the different classes corresponding to the different operating modes of the experimental devices. The employed process has demonstrated exceptional proficiency in classifying the operational states of the analyzed machine and subsequently achieving precise identification and recognition of the health condition, type, and specific faulty element. The future of intelligent mechanical fault diagnosis is very promising, especially in light of the great development in the field of artificial intelligence, this development can keep machines in good working condition for a longer period of time.