A study on attribute selection for job shop scheduling problem

Attribute selection is an effective approach to improve the inference efficiency of data-based scheduling strategies system that many researchers have studied based on computational intelligence methods. Comparing to computational intelligence methods, concept lattice has the advantages in attribute selection protentially. In this paper, attribute selection for production line in job shops based on concept lattice is studied and applied in the neural network (NN) scheduling system. Firstly, owing to the many-valued characteristic of production line attributes, the method of many-valued formal context converts to single-valued formal context is given. Then, the attribute feature is discussed and a concept lattice reduction method for production line attribute selection is proposed to obtain the key production line attributes. Finally, the key attributes are used as the input of neural network scheduling system which can generate optimal scheduling strategies for job shop scheduling problem. The experimental results show that the proposed scheduling system is effective in terms of various performance criteria.