LAUSR

A multi-objective scheduling algorithm with self-evolutionary feature for job-shop-like knowledgeable manufacturing cell (JSKMC) is proposed in this paper, targeting such scheduling issues as make-span, mean complete time of tasks, total tardiness of tasks, number of tardy tasks and the maximum tardiness. Four matrixes are designed to represent the scheduling model of JSKMC. Properties of the key arcs of tasks are discussed and it is found helpless to seek a better solution by reversing the direction of the middle key arcs of tasks. A simplified neighborhood is then established whereby the number of feasible solutions to be searched for is greatly reduced. Based on the above a multi-objective scheduling algorithm with self-evolutionary feature for JSKMC is proposed. Adaptive heuristic critic method is adopted in the algorithm, whose associate search element (ASE) module is designed to select the appropriate action for acquisition of a better solution in the next step by using the knowledge obtained from learning; such an ability of this module can be improved progressively with the increasing training. A scheduling algorithm based on ASE is developed, in which a Pareto archive is embedded to obtain the Pareto optimal solutions. Numerical simulation results confirm the strong ability of the proposed algorithm to home in on the optimal solution by self-evolution via learning.