Distributed Contribution-Based Quantum-Behaved Particle Swarm Optimization With Controlled Diversity for Large-Scale Global Optimization Problems

Many cooperative coevolution optimization algorithms have been proposed recently for solving large-scale global optimization problems. These algorithms first decompose a large-scale global optimization problem into several subproblems, each with a specific number of decision variables, and then optimize the subproblems separately. However, if computing resources are not reasonably allocated to subproblems, computational resources may be wasted. In this paper, we propose a distributed contribution-based quantum-behaved particle swarm optimization with controlled diversity (DC-QPSO) for large-scale global optimization problems. According to the level of optimized contribution of each subproblem, the computing resources are reallocated automatically in each stage, guaranteeing that subgroups with more contribution get more computational resources. Moreover, a parallel diversity control strategy is proposed to enhance the capability of finding better solutions to problems. CEC’2010 and CEC’2013 benchmark function suits are selected to test the performance of the proposed algorithms. The experimental results demonstrate the better performance of the proposed DC-QPSO on the benchmark functions, compared to other state-of-the-art large-scale global optimization algorithms.