Detection and Localization of Faults in System Dynamics by IMM Estimator.pdfVIP

Detection and Localization of Faults in System Dynamics by IMM Estimator Ludmila Mihaylova 1 Emil Semerdjiev1 X. Rong Li Bulgarian Academy of Sciences University of New Orleans Central Laboratory for Parallel Processing Department of Electrical Engineering ‘Acad. G. Bonchev’ Str., Bl. 25-A, 1113 Sofia, Bulgaria New Orleans, LA 70148 Phone: (359 2) 979 66 20; Fax: (359 2) 707 273 Phone: 504-280-7416, Fax: 504-280-3950 E-mail: lsm@bas.bg, signal@bas.bg E-mail: xli@uno.edu Abstract—The cost-effective Interacting Multiple- known as a “drift.” Test examples demonstrating the Model (IMM) algorithm is applied for rapid and efficiency of the approach are given. reliable fault detection and localization in system dynamics. The paper also presents a new IMM ap - 2. IMM Estimator for Fault Detection proach for system parameter drift detection and e s- timation based on augmented state models. Consider a system S composed of independent (non-interacting) subsystems S , i 1,2,...,q con- i Key Words: multiple models, fault detection, IMM nected in parallel (Fig. 1), with state vectors xi , i 1,2,...,q . In