Machine learning for engineering design toward smart customization: A systematic review

In today's manufacturing industry, companies are striving to provide customized products to maintain competitiveness. The challenge of design customization lies in the company's ability to balance product variety, responsiveness, and cost-effectiveness simultaneously. Today, the large volume of data in tandem with powerful computation capabilities has made machine learning a promising technology to address various challenges in engineering design, leading to new opportunities for customization. However, few efforts have been devoted to systemically reviewing these new methods, nor to assessing how they are aligned with customization. Against this background, this article presents a systematic literature review on machine learning for engineering design from the customization perspective. A thorough search of relevant works resulted in a total of 116 most relevant articles, based on which, different machine learning applications are mapped to corresponding design stages of an engineering design process. The potential and advantages of machine learning for fulfilling different customization requirements are discussed. Finally, some promising directions for future investigation are outlined.