In this paper, we propose a hybridization of electromagnetism-like (EM) algorithm and particle swarm optimization (PSO) method to design recurrent fuzzy neural systems for nonlinear control. The hybrid algorithm (called modified EMPSO) combines the advantages of EM and PSO algorithms to enhance the performance of optimization. The main modification from EM algorithm is the random neghborhood local search is replaced by PSO algorithm with an instant update strategy. Each particle’s velocity is updated instantaneously and it provides the best information for other particles. Thus, it enhances the convergence speed and the computational efficiency. Simulation results of nonlinear systems control and two-degree-of-freedom helicopter system are shown to illustrate the modified EMPSO has the ability of global optimization, faster convergence, and higher accuracy.
|Number of pages||16|
|Journal||International Journal of Computational Intelligence in Control|
|State||Published - 2020|
- Electromagnetism-like algorithm
- Fuzzy neural system
- Nonlinear control
- Particle swarm optimization