T-S Fuzzy Model Identification of Hybrid Differential and Multi Particle Swarm Optimization
An hybrid differential evolution(DE)with multi particle swarm optimization(PSO)(HDEMPSO)is proposed for modeling T-S fuzzy system.The premise parameters and consequences parameters are encoded to-gether and identified.To avoid the premature convergence and low convergence speed of the PSO that effect the accuracy and speed,the proposed algorithm partitions the swarm into several subswarms,Each sub-swarms evolves independently and adjusts the inertia weight adaptively according to the fitness value of parti-cles.The best of each particle in subswarms is updated by DE to further enhance the global search ability.The strategy of randomly selection subswarm's best particle replaced by global particle is used to transfer the information among all the sub-swarms and accelerate the convergence speed.The simulation results for typical nonlinear and chaotic systems show that the T-S model with hybrid difference and multi particle swarm optimization has higher identification accuracy.
NETL
NSTL
万方数据
