On the combination of fuzzy models

Abstract

The combination of fuzzy models could be an effective way to improve system performance. This text proposes a fuzzy approach to the combination of fuzzy models, i.e., the different fuzzy models are combined using a fuzzy rule-based model. The combining fuzzy model is identified using an algorithm that is stable towards disturbances. The combination approach provides simultaneously the benefits of the individual components and thus improves overall performance. The combination scheme could be used to resolve the issue of choice of performance deciding parameters (e.g. learning rate).