Optimization of spinning processes in textile manufacturing using reinforcement learning

In smart manufacturing, the integration of data-driven reinforcement learning algorithms is crucial for enhancing decision-making processes to optimize manufacturing within production systems. The textile industry, characterized by its dynamic quality-affecting parameters, demands the seamless assimilation of smart manufacturing technologies. This study introduced a reinforcement learning algorithm that provided valuable insights and recommendations for the optimization of spinning processes. The proposed algorithm ensured the quality of textiles with enhanced strength and elongation characteristics. The Q-learning algorithm, which was devised by leveraging insights from a previously established regression model for predicting textile strength and elongation, served as the foundational element for reward calculations within the Q-learning algorithm. In contrast to conventional experimental methods, this study adopted a data-driven approach that reduced production costs and time and cultivated an enhanced understanding and application of intelligent manufacturing techniques in both academic and industrial settings. The proposed methodology is expected to guide the industry in adopting intelligent algorithms that shape the next generation of textile manufacturing processes.