Dynamic hybrid swarm intelligence approach to generate fuzzy rules
Haixiao Chi, Jin Gou, Cheng Wang, Zongwen Fan · 2016
An important application field of swarm intelligence algorithms is fuzzy rule acquisition. However, their limitations are showed in two aspects. On one hand, it takes a long process to create fuzzy rules during the iterations; on the other, the swarm intelligence algorithms obtain local optimal solution at times. To overcome these disadvantages, a dynamic hybrid swarm intelligence approach is proposed to generate fuzzy rules from data. In this approach, the dynamic adjustment strategy accelerates convergence rate of the swarm intelligence algorithms, meanwhile, the hybrid particles are introduced to avoid being trapped in the local optimum. We adopt the Chinese Longitudinal Healthy Longevity Survey data to prove the effectiveness of the proposal. According to the experiments, the dynamic hybrid swarm intelligence approach provides a competitive results comparing with the differential evolution algorithm, particle swarm optimization algorithm, social emotional optimization algorithm, and Wang Mendel method.