Two-level balancing multi-objective algorithm for trapezoidal type-2 fuzzy flexible job shop problems

Uncertainty remains a critical issue in realistic applications in many fields. However, there is little literature considering trapezoidal interval type-2 fuzzy set (TIT2FS) in scheduling problems. To fill this gap, we propose a two-level balancing multi-objective optimization algorithm for a distributed flexible job shop scheduling problem (DFJSP), wherein two typical constraints are considered: the TIT2FS processing time and the transportation resource limitation. Two objectives are to be minimized: the TIT2FS fuzzy makespan and energy consumption. A two-level evolution mechanism is employed. The first level comprises two populations to optimize the two respective objectives, while the second level is composed of two populations to perform the convergence and diversity balancing tasks. Next, an efficient problem-specific initialization method with several heuristics is presented. Subsequently, different types of mutation and crossover operators are proposed under the consideration of problem knowledge. To further improve the performance of exploration and exploitation abilities, we present a balancing enhanced local search method. Finally, we performed detailed comparisons with four efficient algorithms. The results show that the two-level optimization algorithm was efficient in both convergence and diversity abilities, indicating that our algorithm can efficiently achieve a balance between these abilities.