Mathematical modeling and multi-attribute rule mining for energy efficient job-shop scheduling

Abstract Manufacturing industry accounts for about one-third of the world's total energy consumption (TEC). This study aims to develop a novel mixed-integer mathematical model to represent the direct energy consumption of machines and indirect energy consumption on a shop floor. In comparison with traditional modeling methods, this paper proposes an effective gene expression programming-based rule mining (GEP-RM) algorithm to generate dispatching rules automatically. This method consists of three attributes that have significant impacts on the TEC of a manufacturing process. In addition, diversified rule mining operators with self-learning are designed to ensure population diversity and convergence. Moreover, a perturbation trigger mechanism for reconstructing rules is introduced to avoid being trapped into a local optimum. An unsupervised learning algorithm is achieved by setting the evolution direction with global best and current worst in order to mine the value of the substantial historical data. Experimental results have shown that the proposed multi-attribute rule mining approach outperforms other dispatching rules in terms of energy saving and production efficiency.