Tuning fuzzy systems parameters with chaotic particle swarm optimization
I. A. Hodashinsky, Marina Bardamova · Journal of Physics Conference Series · 2017
In this paper, we investigate the application of chaotic particle swarm optimization (PSO) to fuzzy system parameter estimation. Unlike traditional PSO, chaotic PSO improves search capabilities of particles using chaotic transformations of random coordinates. Different mapping functions were investigated to generate chaotic transformation sequences. The efficiency of the algorithm was tested on approximation problems originating from data sets of the KEEL repository. The experimental study compared several existing techniques for fuzzy systems identification. The results indicate that chaotic PSO is a competitive approach to fuzzy system identification able to create fuzzy systems of high precision at an acceptable set of rules.