The detection of change points in a switched system

Abstract

Change point analysis is a growing subject. The earliest application of change point analysis was in engineering quality control, but in recent years change point analysis has been widely used in other fields, such as clinical trials and economics. In this thesis, first I discussed why analysis epidemic change point is more important than the other kinds of change points. And this thesis only focused on epidemic change point, which only switching from two sub-systems. Compares to the other kinds of change points, epidemic change point has more applications in real life, especially in the fields of quality control and medical trial. I will discuss the detection of change points in two different methods. First method is using the property of the sub-matrix of Hankel matrix along with the characteristics polynomial to detect change points. Because of the limitation of this method, the second method discussed in the paper is based on information criterion. And mainly focused on the stability of the hypotheses testing.