OptiFel: A Convergent Heterogeneous Particle Swarm Optimization Algorithm for Takagi–Sugeno Fuzzy Modeling

Data-driven design of accurate and reliable Takagi-Sugeno (T-S) fuzzy systems has attracted a lot of attention, where the model structures and parameters are important and often solved in an optimization framework. The particle swarm optimization (PSO) algorithm is widely applied in the field. However, the classical PSO suffers from premature convergence, and it is trapped easily into local optima, which will significantly affect the model accuracy. To overcome these drawbacks, we have developed a new T-S fuzzy system parameters searching strategy called OptiFel with a heterogeneous multiswarm PSO (MsPSO) to enhance the searching performance. MsPSO groups the whole population into multiple cooperative subswarms, which perform different search behaviors for the potential solutions. We have found that the multiple subswarms strategy proposed in this paper is greatly helpful for finding the optimal parameters suitable for the subspaces of the T-S fuzzy model. Our theoretical proof has also demonstrated that the cooperation among the subswarms can maintain a balance between exploration and exploitation to ensure the particles converge to stable points. Experimental results show that MsPSO performs significantly better than traditional PSO algorithms on six benchmark functions. With the improved MsPSO, OptiFel can generate a good fuzzy system model with high accuracy and strong generalization ability.