Cooperative hybrid EA for large-scale flexible job shop scheduling

Flexible job shop scheduling problem (fJSP), which belongs to the classic combinatorial optimization problem, is difficult to solve with exact methods. Evolutionary algorithm (EA) has been widely used for dealing with fJSP in recent years. Large-scale flexible job shop scheduling problem with high complexity is of great importance in a real industrial production environment and indicates an advanced requirement for traditional EAs. In this paper, we propose a cooperative hybrid EA (ChEA) to solve large-scale fJSP with the objective of minimizing the makespan. fJSP with significantly complex encoding and decoding procedure is simulated as a two-stage random key-based representation. An effective set-based random grouping paradigm is used to decompose the variables space and solution space into small scale ones, achieving cooperative co-evolution optimization. We employ the particle swarm optimization (PSO) based on Gaussian distribution and local best individual as the evolution algorithm. Local search of moving two operations on the critical path is adopted to enhance exploitation. Numerical experiments carried out on large-scale instances get competitive performances compared with state-of-the-art algorithms.