الفهرس 
FAULTS CLASSIFICATION AND MODELINGOnly 14 pages are availabe for public view
 |  | from | 129 |  |  |
| | | from | 129 | | |
Abstract
Since there is no system in the real world that can work perfectly at all time, faulttolerant control systems (FTCSs) have become a critical issue especially for modern
complex technological systems that rely on sophisticated control functions to meetincreased performance requirements such as reliability, survivability, maintainabilityavailability, cost efficiency, safety and quality. The fault tolerant control systemFTCS) is defined as a control system that possesses the ability to accommodatesystem component faults automatically. Recently, fault tolerant control (FTCmethods are categorized into passive and active approaches. In passive FTCcontrollers are designed to be robust against a class of presumed faults. Thereforepassive FTC approach needs neither FDD schemes nor controller reconfiguration and
the resulting controller usually has a fixed structure and parameters. In active FTCcontrollers employ both system restructuring and control reconfiguration to improvesystem performance in the presence of faults. Crucially to prevent significantperformance degradation or damages to the system, it is necessary to detect andidentify the possible faults in the system as early as possible. Accordingly, faultdetection and diagnosis (FDD) scheme with high sensitivity to faults and ability toprovide precise and the most up-to-date information about the system as soon aspossible after the fault occurrence is required. This thesis adopts an integrated activefault tolerant control system (AFTCS) design relies on a fault detection and diagnosisFDD) process to monitor system performance, detect abnormal conditions throughresiduals generation algorithm, isolate the faulty component(s) in the system andestimates the effectiveness factor relating to the detected fault. A model-based faultdetection and diagnosis (FDD) technique using a bank of constrained Kalman filter
estimators have been designed in order to detect, isolate, and estimate faults, andprincipally to distinguish between sensor and actuator faults relying on detailedprecise knowledge of system dynamical model. An explicit fault estimation algorithmis derived to estimate the effectiveness factor of a detected faulty sensor or actuatoriiAccordingly, a systematic approach to an on-line controller reconfiguration design forthe post-fault system is activated to recover both transient and steady-state
performance in spite of fault occurrences. A reconfigurable state feedback controllerwith reference input design technique is proposed using singular value decomposition
(SVD) based solution of eigenstructure assignment (EA) to guarantee the stability ofthe reconfigured system by recovering the system eigenvalues and recovers the prefaultsystem performance to the maximum extent by placing the eigenvectors as closeto those of the original system as possible. The effectiveness of the proposed fault
detection and diagnosis (FDD) scheme has justified by simulation result on thebenchmark Three-Tank System with simulated actuator and sensor faults. Theeffectiveness of the demonstrated approach of developed active fault tolerant controlsystem (AFTCS) with combination of both control reconfiguration algorithm and faultdetection and diagnosis (FDD) scheme has justified by simulation result on thesteering subsystem of the Naval Postgraduate School (NPS) Unmanned UnderwaterVehicles (UUV) with simulated actuator (rudder) faults. The performances of thecontrol reconfiguration approach are emphasized by simulation results of the fault-freecase, the faulty case without reconfiguration and fault accommodation with controllerreconfiguration.