Highlights

Abstract

Keywords

۱٫ Introduction

۲٫ Literature review

۳٫ Problem formulation

۴٫ Methodology

۵٫ Case study

۶٫ Conclusions and future work

Declaration of Competing Interest

Acknowledgments

References

Abstract

Multi-objective optimization, such as quality, productivity, and cost, of the textile manufacturing process is increasingly challenging because of the growing complexity involved in the development of textile industry in the upcoming big data era. It is hard for traditional methods to deal with high-dimension decision space in this issue, and prior experts’ knowledge is required as well as human intervention. This paper proposed a novel framework that transformed the textile process optimization problem into a stochastic game, and introduced deep Q-networks algorithm instead of current methods to approach it in a multi-agent system. The developed multi-agent reinforcement learning system applied a utilitarian selection mechanism to maximize the sum of all agents’ rewards (obeying the increasing ε-greedy policy) in each state, to avoid the interruption of multiple equilibria and achieve the correlated equilibrium optimal solutions of the textile process. The case study result reflects that the proposed MARL system can achieve the optimal solutions for the textile ozonation process, and it performs better than the traditional approaches.

۱٫ Introduction

The textile manufacturing process adds value to fiber materials by converting the fibers into yarns, fabrics, and finished products [1]. Under the arousing global competition, textile companies have to face the challenges of cost reduction and performance improvement. There is a growing public concern on the environment which imposes bounds to the textile manufacturers on the exploitation of power, water and resources. The future development of textile manufacturing relies heavily on product customization and shortened manufacturing cycles since the distributors and consumers are increasingly looking for flexible capacity sensitive to demand variability. To deal with the high degree of variability in materials, processes and parameters, the manufacturers traditionally conduct trial and error, and lean on the expertise and experience [2]. There is a strong need to develop innovative methods to improve the textile manufacturing process.

Since textile manufacturing consists of a very long value chain of processes from raw materials to finished products (a brief example is provided in Fig. 1), the combinations of processes and parameters at different stages could be stochastic and immense when factors of the targeted performance vary in any respect [3–۵]. And because of the number of factors such as increasing component (or product) complexity, it is difficult to obtain the optimal scenario of a textile manufacturing process. Meanwhile, the performance of the textile process is always governed by a few criteria and the quality of their significance with an overall objective is different [6]. Thus the optimization problems in this domain always take multiple objectives into account. It is very challenging for the simultaneous optimization of multiple targets in a textile production scheme from high dimensional space.

Scholars tended to employ mathematical programming methods and meta-heuristic algorithms to overwhelm textile manufacturing process optimization problems. Krishna et al. [7] utilized dynamic programming models to find the optimal maintenance policy of sewing machine and to decrease their costs in the textile industry. Majumdar [8] applied linear programming to maximize the overall profit of functional clothing production, and applied goal programming to optimize two conflict objectives, namely ultraviolet protective property and air permeability, of the functional clothing.