A rules‐firing strength‐based neuro‐fuzzy observer for information‐poor systems

Abstract

In this paper, a neuro-fuzzy observer (NFO) is proposed for estimating the unmeasured states of an information-poor system by relaxing the strictly positive real condition (without using filtered fuzzy basis function (FBF) and filtered output estimation error) and without using high-gain terms. To recover the performance of the observer in the absence of high-gain terms, a concept of weighted fuzzy rules (or strengthened FBF) is proposed. The weighted fuzzy rules are then used to propose the concept of weighted function approximation which is then utilized in the design of NFO to estimate the unknown dynamical function. For reducing computational power and avoiding over-tuning of the weights, a concept of relay-switching is also introduced in the design of the NFO. The stability analysis of the proposed NFO is also presented using Lyapunov approach and the effectiveness of the proposed scheme is demonstrated through a simulation example.