A Neural Network-Based Multiplicative Actuator Fault Detection and Isolation of Nonlinear Systems

Abstract

The problem of fault detection and isolation/identification (FDI) of nonlinear systems using neural networks is considered in this paper. The proposed FDI approach employs recurrent neural network-based observers for simultaneously detecting, isolating and identifying the severity of actuator faults in presence of disturbances and uncertainties in the model and sensory measurements. The neural network weights are updated based on a modified dynamic backpropagation scheme. The proposed FDI scheme does not rely on the availability of full state measurements. In most works in the literature the fault function acts as an additive term on the actuator, whereas in this work the fault acts as a multiplicative term. This will make the formal stability and convergence analysis of the overall FDI scheme nontrivial and challenging. Our stability analysis considers the presence of plant and sensor uncertainties through the use of Lyapunov's direct method with no restrictive assumptions on the system and/or the FDI algorithm. The performance of our proposed FDI approach is evaluated through simulations that are performed for two case studies, namely FDI of 1) reaction wheel type actuators that are commonly utilized in the attitude control subsystem (ACS) of a satellite and 2) actuators in a two-link flexible joint manipulator.Index Terms-Fault detection and isolation, multiplicative actuator faults, nonlinear systems, recurrent neural networks.

Adaptive tracking control of a nonlinear teleoperation system with uncertainties in kinematics and dynamics

“…To deal with the uncertain kinematics and dynamics preceded by robotic systems, some interesting results have been presented in the literature. [12][13][14][15] Cheah et al 12 developed an adaptive Jacobian controller for the tracking control of a manipulator with kinematic and dynamic uncertainties. Moreover, Wang and Xie 13 further proposed an adaptive inverse dynamics controller to ensure the uniform performance of robot over the entire workspace.…”

Fault Mechanism Analysis for Manufacturing System Based on Catastrophe Model

“…Datadriven fault analysis rests on either an explicit mathematical model derived from prior knowledge or a reasoning mechanism derived from experience. It uses different types of data mining technology to extract and classify fault features in acquired vast operating data [4], which include signal processing [5,6], statistical analysis [7,8], and early quantitative artificial intelligence methods [9]. With increasing degrees of automation and intelligence of industrial equipment, along with the development and widening of application spectrum of related advanced technologies [10,11], data began to grow exponentially.…”

Actuator fault detection and isolation of combined system based on space geometry theory

“…A methodology using a neural network observer strategy to solve the actuator FDI problems for nonlinear systems is presented, and the neural network weights are updated based on a modified dynamic back propagation scheme [3]. For satisfying the so-call observer matching condition that restricts practical applications of algebraic unknown input observers for fault detection and isolation, reference [4] presents a scheme to design a fault detection observer that decouples the effect of mismatched unknown inputs on the residual for linear systems.…”