A Novel Robust Fault Detection Scheme of Lipschitz Nonlinear Systems Using Combination of Bond Graph and Observer

TL;DR

This paper introduces a robust fault detection scheme for nonlinear Lipschitz systems that combines Bond Graph modeling with a novel observer-based residual method, enhancing noise robustness and disturbance rejection.

Contribution

It proposes a new Error-based Analytical Redundancy Relations (EARRs) method integrated with Bond Graphs and a nonlinear observer for improved fault detection in Lipschitz systems.

Findings

Enhanced fault detection accuracy in simulations

Improved robustness against noise and uncertainties

Demonstrated effectiveness on a manipulator model

Abstract

This paper deals with a new robust fault detection (FD) scheme for nonlinear Lipschitz systems wherein a robust nonlinear observer is used in combination with the Bond Graph (BG) method. In order to improve the efficiency of the classical Analytical Redundancy Relations (ARRs) FD scheme based on the BG method, a new form of the ARRs for the nonlinear Lipschitz systems is presented. This new form of the residuals is based on the output estimation error of the observer and is called Error-based Analytical Redundancy Relations (EARRs). The robustness against disturbances and parametric uncertainties in the FD system is achieved by the proposed method. Next, the integral form of the EARRs for the nonlinear Lipschitz system is presented that is robust in the sense of measurement noises as well. The BG model of a single link manipulator with revolute joints actuated by a DC motor is derived and the efficiency of the proposed method in comparison to the classical ARR method is shown in the simulation results.