We propose a useful adaptive control method on fuzzy control using a fuzzy associative memory system and LVQ; learning vector quantization. LVQ, which is proposed by T. Kohonen, can automatically cluster the input space on neural networks. We use LVQ to automatically create membership functions of fuzzy rules at on-line computing without teaching signals. We realize a new fuzzy adaptive control which uses three elements; (1) membership functions created by LVQ, (2) fuzzy model-type fuzzy rules, and (3) a fuzzy associative memory system. The fuzzy associative memory system is useful for fuzzy inference on this fuzzy adaptive control. The fuzzy model-type fuzzy rules are useful for this fuzzy adaptive control because of high representation faculty. On an application, we show this fuzzy adaptive control is useful for a non-linear system with variable parameters.
J-STAGEがリニューアルされました! https://www.jstage.jst.go.jp/browse/-char/ja/