Minimum-energy neural-fuzzy approach for current/voltage-controlled electromagnetic suspension system

Abstract

In this paper, the electromagnetic suspension system is modeled as a neural-based linear T-S fuzzy system, and then the optimal fuzzy control design scheme is proposed to control the current and voltage-controlled system with minimum current and voltage consumption, respectively. The proposed linear self-constructing neural fuzzy inference network is a six layer neural network (linear SONFIN) modified from the well-known SONFIN network, which can construct a linear T-S fuzzy model of the system just by the input and output (I/O) information. Based on the linear T-S model, we can construct the optimal fuzzy control scheme to efficiently regulate the highly nonlinear, complex and uncertain electromagnetic suspension system to the equilibrium state.