Solving Dynamic Optimization Problems Based on an Improved Clustering Quantum-Behaved Particle Swarm Optimizer

摘要

Solving dynamic optimization problems (DOPs) has become a hot research area in recent years. In this paper, an improved Quantum-behaved particle swarm optimizer (QPSO) with hierarchical clustering technique is proposed to solve DOPs in terms of the dynamics and complexity. By studying the trajectory of particles and the effect of local attractor in QPSO, Gaussian distribution is used to sample the local attractor in order to alleviate premature convergence and enhance the global search ability. Two variants of the proposed Gaussian distributed are also studied in this paper. For the hierarchical clustering technique, each particle is regarded as one cluster in the initial stage. With the evolutionary procedure, the clusters merge into new sub-swarms when the distances between them are less than a threshold, which is used for tracking the movement of peaks. The convergence check, overcrowding check, and overlapping check are appointed during the clustering procedure in order to keep swarm diversity. The performance of the improved QPSO with gaussian sampling and clustering technique is evaluated on a set of benchmark functions and compared with some state-of-art algorithms. Experimental results show the efficiency of the proposed approach.