Recurrent fuzzy controller design by two-stage genetic algorithm with local and global mapping searches
Chia‐Feng Juang · 2003
A recurrent fuzzy controller design in reinforced learning environment using two stage genetic algorithm (TSGA) with concurrent local and global mapping searches is proposed. For the local mapping search stage, it searches through the well performed local rules. In this stage, each individual represents only a fuzzy rule. As to the global-mapping search stage each individual represents a whole fuzzy network. The stage determines which local rules should be combined together to achieve a good fuzzy network. Finally, the performance of TSGA is verified from simulation comparisons.