Recursive Subspace identification for time-varying continuous-time stochastic systems via distribution theory

This paper proposes a recursive subspace identification method for time-varying continuous-time stochastic systems based on distribution theory. By using the random distribution theory, the time-derivative of stochastic processes is described and the input-output matrix equation is obtained. Moreover, we reduce the storage cost and computation burden by keeping the size of input-output data to be constant. Further, the system model is obtained. The simulation results show the validity and accuracy of the proposed method.