An Interactive Simulated Annealing Multi-agents Platform to Solve Hierarchical Scheduling Problems with Goals

In this paper, an Interactive Simulated Annealing Multi-agents (ISAM) platform is developed to solve hierarchical multicriteria scheduling problems with goals, considering four criteria. Solving scheduling problems with this number of criteria is missed in the literature. The proposed algorithm is composed of three phases: at the first one, four agents are created to independently optimize four criteria with different random initial solution and initial configuration. This phase is achieved by proposing to the decision maker one value for each criterion. At the second phase, for each criterion a goal is fixed by the decision maker. At the third one, two agents are launched with adaptive memory to hierarchically minimize the deviations of the solutions from the goals. To fully evaluate the effectiveness of the proposed ISAM approach, we test it on different problems: a) lexicographic goal programming problems with continuous variables to prove its platform efficiency; b) single machine total weighted tardiness problems to show its robustness to solve NP-HARD scheduling problems; c) single machine problems to hierarchically solve four criteria with goals.