Continuous-time subspace system identification method

Abstract

We analyze practical problems of discrete-time subspace system identification methods from the control point of view and develop a new continuous-time subspace system identification method. We use a new transform to obtain the derivative information numerically from the measured process output and the process input. Using the transform, we convert the continuous-time process model composed of differential equations into corresponding algebraic equations. We estimate the process order by inspecting singular values, and subsequently, we determine the continuous-time state space model using the least-squares method. We compare the proposed method with the discrete-time subspace system identification method for a small sampling period.