System Identification and Indirect Inverse Control Using Fuzzy Cognitive Networks with Functional Weights

Abstract

A Fuzzy Cognitive Network (FCN) is an operational extension of a Fuzzy Cognitive Map (FCM) which assumes, first, that it always converges to equilibrium points during its operation and second, it is in continuous interaction with the system it describes and may be used to control it. In this paper we show that the conditions that guarantee the convergence of the FCN may lead to a special, yet very powerful, form of the network that assumes functional interconnection weights with excellent system approximation abilities. Assuming that the plant is unknown it is initially approximated by a FCN and a procedure for adaptive estimation of its functional weights is proposed that guarantee approximation error convergence to zero. The FCN is then used for the Indirect adaptive Inverse Control of a plant. The methodology is tested on a coupled two-tank system.