Fault Detection Diagnosis and Fault Tolerant Control of Descriptor Systems

Abstract

methods yield system representation in such form and it represents a larger class of systems. newlineDescriptor systems or Differential Algebraic Equations(DAEs) are different from ordinary newlinedifferential equations(ODEs) which limit ODE methods to be applied directly to DAEs. newlineMajority of works focus on numerical analysis, solution, stability and control of descriptor newlinesystems. However, prior work in the domain of Fault diagnosis(FD) and Fault Tolerant newlineControl(FTC) for descriptor systems is sparse in comparison to work done for conventional newlinesystems. The main objective of the thesis is to develop approaches for state estimation newlineand fault diagnosis for a class of descriptor systems. The study intends to accomplish state newlineestimation, fault detection, diagnosis, and fault tolerant control for descriptor systems. newlineAt the outset, the existing approaches for state estimation of nonlinear descriptor systems newlineare evaluated. Further, based on that, novel filters employing evolutionary approaches newlinewithin the Extended Kalman filter framework are proposed and demonstrated on benchmark newlinenonlinear descriptor systems. The performance of the proposed filters is evaluated newlinein a quantitative manner which shows improvement in algebraic state estimation performance. newlineFurther, this study addresses the issue of fault detection by employing computationally newlineinexpensive residual evaluation methods. newline newline