Manufacturing Process Parameter Optimization: A Comprehensive Review and Research Directions

Abstract

Manufacturing industries face significant challenges in minimizing the scrap generated during production processes. Any measurable input element in the manufacturing process, such as temperature, raw material quality, or wire feed rate, along with any transformation activity within the process that affects the output, is considered a process parameter. Optimizing these process parameters through various techniques can reduce scrap rates and enhance product quality. This paper reviews and synthesizes the available literature on manufacturing process parameter optimization. The survey emphasizes key manufacturing processes, including shaping, forming, machining, joining, finishing, and chemical processing. It categorizes process parameter optimization approaches into online and offline methods, revealing that online optimization currently accounts for only 6-7% of the research focus. A retrospective analysis underscores the need for further research in online optimization for dynamically changing environments. The previous review from Weichert et al. (2019) focused narrowly on machine learning methods (2008–2018) for quality improvement.  Likewise, Panzer & Bender (2022) examined only deep reinforcement learning in production. In contrast, this review covers a broader algorithmic map across all major manufacturing processes (shaping, forming, machining, etc.). It uniquely quantifies the split between offline and online studies. The paper concludes by discussing the challenges in implementing solutions for process parameter optimization and calls for additional academic and industrial research, particularly through advanced machine learning techniques, to optimize manufacturing process parameters.