A Multi-Objective Fuzzy Job Shop Scheduling by Extremal Optimization

Job shop scheduling deals with distributions between jobs in order to find a schedule with minimum possible time. To formulate job shop scheduling problems, various factors such as the activity processing time and due date for delivering jobs are often ambiguously known to the analyst. In these situations, the use of fuzzy parameters and multi-objective goals based on fuzzy knowledge seems necessary. Since this problem is non polynomial, we propose a method based on extremal optimization in order to select and modify less valuable but more probable activities. This method leads to decrease the number of less valuable activities and to increase the number of activities with the same value. In such cases, any partial change in scheduling may produce big changes. For instance, it leads to escape from local optima and move toward global one. Lots of simulations on and experimental data, demonstrate proper convergence rate and good results in comparison with other methods. The correctness of given results and proposed algorithm have been proven by means of axiom of convergence, statistical t-test t and possibility error (F).