Study On Attributes Selection And Scheduling Rules Mining Of Job Shop Scheduling Problem

Job shop scheduling problem is a typical production scheduling problem, with characteristic of dynamic, uncertainty and computational complexity. In last decades, researchers have proposed a number of algorithm for solving job shop scheduling problem. Scheduling rules with low computational complex-ity and easy to understand, is widely used in the real-world manufacturing plants. However, studies have shown the state of production line has a great impact on the performance of scheduling rules and single dispatching rule cannot adapt to the changing production environment. It is necessary to design the scheduling rule-based adaptive scheduling system. Meanwhile, with the development of information technology, manufacturing system has accumu-lated a lot of information relevant to scheduling and data-based production scheduling method attracts widespread attention. This paper focuses on the attribute selection and scheduling rules mining of data-based scheduling method in job shop scheduling problems. The main contents are as follows:In view of the need of samples for attribute selection method and scheduling rules mining algorithm, a simulation platform to extract optimal scheduling rule for job shop scheduling based on Multi-Pass is established. The job shop scheduling problem ft10 for the study, the platform is built in Plant Simulation based on Multi-Pass simulation. By running simulation, samples for attribute selection and scheduling rules mining are constructed.Two attribute selection methods is studied to solve the problem that a large number of irrelevant or redundant attributes can affect the performance of scheduling rule mining algorithm in the production environment. One is the samples of scheduling attributes are mapped to the many-valued formal context in concept lattice and the feature attributes are selected according to the importance of attributes. The other is firstly to convert the many-valued formal context to single-valued formal context and then the feature attributes selected by using concept lattice reduction theory.A scheduling rules mining method based on attribute selection for job shop scheduling problem is study for dynamic production environment. Based on neural network, the proposed method with feature attributes as input improves the mining accuracy of scheduling rules. Meanwhile, the proposed method can generate optimal scheduling rule according to the status of current production environment.