I am gonna find change points in ordinary differential equations modeling where the current parameters cannot simulate the real data set anymore, and need to segment data set based on detected change points so that we can do parameter estimation in each segment to campture the whole dynamic. Is there any hint or reference for this?
I tried some statistical methods for change point detection, but they could not solve my problem because they only consider the real data set to detect change points but in my case the criteria for change has to use model simulation and real data set at a same time.