Date of Award
Spring 2019
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Electrical and Computer Engineering
First Advisor
Schneider, Susan C.
Second Advisor
Yaz
Abstract
In this thesis, two discrete-time control systems subject to noise, are modeled, analyzed and estimated. These systems are then subjected to attack by false signals such as constant and ramp signals. In order to find out how and when the control systems are being attacked by the false signals, several detection algorithms are applied to the systems. This work focuses on actuator attack detection. To detect the presence of false actuator signals, a bank of Kalman filters is set up which uses adaptive estimation and conditional probability density functions for detecting the false signals. The individual Kalman filters are each tuned to satisfy a control system: one of which is the original system and the other of which is the system with a false signal. The use of the bank of Kalman filters to detect actuator attacks is tested in 4 cases; first-order system attacked by a constant or ramp signal and then a second-order system subject to the same types of attack signals. This work shows the bank of Kalman filters can successfully detect the intrusion of false signals for actuator attack by using several different detection algorithms. Simulations show that the false signal is found and detected in all cases.