Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Electrical and Computer Engineering

First Advisor

Jorge L. Aravena


In many dynamic systems, the parameters are subject to jumps at unknown points in time. The jumps may be the result of some change in the operating system, or plant failure. In this work a method has been developed for fast detection of these jumps and estimation of the post-jump values. The proposed method is called the Combined Filter Algorithm. It is based on the conventional Kalman filter. Single-sample hypothesis test is used to determine the presence or absence of jumps. Small jumps, missed by the hypothesis test, are traced by a Random Walk Model Kalman filter. This is made possible by the introduction of a new kind of decision rule, called the Combined Decision Rule. The properties of this filter under correct and incorrect decisions are studied, and mathematical proofs are presented. The intended main application of the algorithm is the detection of power system faults and estimation of steady-state voltages and currents. The proposed algorithm is tested for ARMA models and discrete state models. The algorithm is also tested on a continuous-time system, where a jump occurs in one parameter. The algorithm is applied to detect and estimate the resulting changes produced in an equivalent ARMA model. Simulation results are presented.