A New Approach to Generate Dispatching Rules for Two Machine Flow Shop Scheduling Using Data Mining

This paper introduces a novel methodology for generating scheduling rules using data mining based approach to discover the dispatching sequence by applying learning algorithm directly to flow shop scheduling. Flow scheduling is one of the well-known combinatorial optimization problems. This paper considers the problem of finding schedule for two machine flow shop to minimize the make span using Decision Tree (DT) algorithm. This approach involves pre-processing of scheduling data into an appropriate data file, discovering the key scheduling concepts and representing of the data mining results in way that enables its use for scheduling. The advantages of DT's are that the dispatching rule is in the form of IF-Then else rules, which is easily understandable by the shop floor people. In decision tree based approach, the attribute selection greatly influences the predictive accuracy and hence this approach also considers creation of additional attributes. For two machine flow shop problem, the Johnson's Algorithm (JA) yields optimal solution. Hence the proposed approach is compared with Johnson's algorithm. The results show that both the methods yield the same result. The work is a complement to the traditional methods.