Signal Processing and Machine Learning Techniques for Predictive Maintenance of Rotor Bars in Induction Machine

New technological innovations, such as integrating information technology with physical devices, have emerged due to the industrial revolution 4.0. This has led to the development of intelligent machines, which fall under the Internet of Things (IoT) field. As a result of this integration, industrial processes have become more efficient and productive. In addition, it has reduced maintenance costs and downtime by applying condition monitoring and machine learning algorithms to electrical machines. Despite these advancements, there is still a lack of algorithms for predicting faults in electrical machines, which is an area of ongoing research. This paper presents a comparative analysis of different machine learning models with a combination of the signal spectrum-based machine learning approach for predicting broken rotor bars in induction machines as a potential solution for predictive maintenance. The method is validated by real collected data from a working induction machine. The preliminary comparison based on the training time and the results’ accuracy is presented.