Machine Tools’ Running State Monitoring in the Learning Factory—Based on Machine Learning Method and Vibration Data

Abstract

Acquisition of machine tools’ running state data plays an important role in the process of workshop management and has great potential in the improvement of Overall Equipment Effectiveness (OEE). Therefore, it is necessary to identify and monitor the running state of machine tools. Machine learning, as a branch of Artificial Intelligence (AI) technology, can use readily available data to derive appropriate models and use them to make judgments about new situations, which means that AI technology is also becoming increasingly important in manufacturing. The Learning Factory provides a way for students and researchers to learn and research AI and condition monitoring by simulating an actual industrial environment. Advanced Manufacturing Technology Center (AMTC) Learning Factory at Tongji University (Shanghai, China) has designed an open and scalable training course on AI technology, including teaching and practice. A method based on machine learning technology is also put forward to monitor the running state of machine tools using vibration data. The Maximum Information Coefficient (MIC) algorithm is used to remove the relevant items from the original feature parameters and reduce the dimensionality to get the feature matrix; the classification model Random Forest is used to complete the classification mapping between the feature matrix and the target matrix to achieve the purpose of indirectly monitoring the machine tools’ running state. The Result could serve as a reference for other Learning Factory AI implementations. This article introduces the method of machine tools’ state monitoring and the design of the training course in detail.