Evolutionary Design of Neural Fuzzy Network Controller
- Hybrid design method for a fuzzy neural network controller (FNNC) that controls the two-links brachiation mobile robot (BMR).
- The method combines an improved simple genetic algorithm and the weighted errors balance algorithm with the global and local learning principles. A known "good control strategy" is used as learning data.
- The incremental rulebase developing algorithm, based on the analysis of the local learning criterion, is proposed in order to simplifie the rulebase.
Main stages of the design method
The local learning criterion and the incremental rulbase developing algorithm