Machine Learning Aided Six Sigma: Perspective and Practical Implementation
研究了机器学习如何辅助六西格玛方法，通过循环神经网络、遗传算法和卷积神经网络改进实验设计和控制图，并结合案例展示了在制造业中利用大数据减少变异、提升质量。
Companies are harnessing the power of artificial intelligence to analyze Big Data and improve their operations. On the other hand, Six Sigma has been used for quality improvement since the 1980s. The use of Six Sigma may redefine industrial precision and provide more practical ways to handle Big Data when aided by machine learning practices. In this research, we explore how these two concepts can work together effectively to further enhance productivity, improve quality, and reduce variation. This article focuses on the ability of define, measure, analyze, improve, and control methodology to adapt to this synergy. We suggest the use of three machine learning methods, namely recurrent neural networks, genetic algorithms, and convolution neural networks, as an aiding tools to the design of experiments and control chart that are commonly used in the analyze and control phases. We also provide a case study to demonstrate the use of machine learning aided Six Sigma in a real-world industrial setting. The proposed machine learning aided Six Sigma applied in this research was found to accurately help make use of the Big Data collected from preinstalled sensors in the manufacturing processes. Machine learning can assist in overcoming Six Sigma constraints caused by human mistakes, inadequate information, and restricted analytical skills. Moreover, it can entertain and utilize massive data collected resulted from technological advancement of sensors and Internet of Things.