Introducing Machine Learning in Undergraduate Mechanical Engineering Mechatronics Classes

Abstract

Abstract Machine learning (ML) that has emerged as a new technology for analyzing big data sets to uncover hidden relationships in an increasingly connected world has a great potential to serve as a critical tool for mechanical engineers to develop better devices and to better understand physical phenomena around us. However, compared to the critical importance and impact of ML and increasingly wide adoption of ML in various mechanical engineering fields, very few mechanical engineering undergraduate students receive direct instruction on ML during their undergraduate work. In order to address these concerns, the goal of this research is to improve undergraduate mechanical engineering students’ understanding and use of machine learning. To accomplish this goal, the research team establishes an integrated education framework to effectively introduce ML to junior-level undergraduate ME students through active and authentic learning by providing hands-on activities with mechanical testbeds. The research team incorporated ML into mechatronics lab activities for measuring mechanical vibrations to introduce and practice ML as an alternative method to analyze vibration signals in addition to classical time and frequency domain vibration analysis, which is one of the important subjects in the curriculum of mechanical engineering. The research group first introduced basic concepts of ML and Long Short-Term Memory (LSTM) model which is one of the ML models suitable for predicting one dimensional or time-series data. Students are then tasked to use the vibration signals they measured as input to train an LSTM model that was provided as a pre-filled MATLAB code with blanks for students to adjust. The results of pre- and post-self-assessment survey asked to the students indicate that students have shown great interest in ML, and considered it important for their careers regardless of the pilot ML learning module. On the other hand, after the pilot laboratory, the students’ self-confidence in ML and mechanical vibrations has greatly increased. Furthermore, the interest in mechanical vibrations increased as well, which provided promising opportunities for enhancing students’ learning. This study results enhance our understanding of how exposure to ML affects undergraduate mechanical engineering students’ understanding and attitude toward ML as a non-computing major and their academic achievement in mechanical engineering.