Failure Modeling of Industrial Electric Motors Using Unsupervised Learning Methods

The task of detecting breakdowns helps to improve the reliability of industrial devices through early identification of problems and their timely resolution, which helps to improve the resource and energy efficiency of production as a whole. When modeling breakdowns of industrial electric motors, the following characteristics are primarily taken into account: average load, rotational frequency, power supply voltage, current consumption, leakage current, collected from data from a traditional automated control system and breakdown logs of repair teams of an industrial enterprise. The model for detecting reliability and breakdowns of electric motors in the work is built on an anomaly detection algorithm using unsupervised machine learning methods. Its effectiveness has been successfully tested using many evaluation metrics. It can be used for a range of tasks and applications in equipment management and maintenance, visualization, forecasting and management decision support.