INTEGRATION OF DATA SCIENCE INTO THERMAL-FLUIDS ENGINEERING EDUCATION

To improve the integration of data science into thermal fluids education, a technical elective course is developed to introduce a wide range of machine learning and deep learning algorithms to engineering students, including principal component analysis, multiplayer perceptron, convolutional neural networks (CNN), long short-term memory (LSTM) networks, reinforcement learning (RL), generative algorithms (GA), and generative adversarial networks for mechanical engineering applications, including visualization-based physical quantity predictions, dynamic signal classification, and prediction, data-driven control of dynamical systems, surrogate modeling, dimensionality reduction, among others. The lectures cover the fundamental concepts and examples of developing machine learning models using Python and MATLAB. To facilitate students' practice of applying data science in solving mechanical engineering progress, this course has touchpoints in several key areas of mechanical engineering, including fluid mechanics, heat transfer, materials science, design, and dynamics/control. Twenty-five students, including seven undergraduate and eighteen graduate students, took the course and the outcomes are very fruitful and encouraging. A variety of data science algorithms have been leveraged to solve mechanical engineering research problems, including generative designs of air-cooled heat sinks, Gaussian process regression for battery lifetime prediction and femtosecond laser manufacturing parameters, GA for two-phase cooling heat exchanger design, coupled PCA and LSTM for microcapsule deformation prediction, CNN for boiling regime classification and laser-manufactured textures classification, bidirectional recurrent neural networks basecalling of DNA and RNA sequences, coupled PCA and MLP porous medium morphology classification, RL for the control of soft robotics. The course has so far led to 25 student-centered machine learning projects, two conference papers, and an Honors thesis.