Decentralized multivariate identification of interconnected systems by a stochastic subspace method

In this work an identification methodology for decentralized multivariate identification of serial interconnected systems by a stochastic subspace method is proposed. The global system identification is made through the individual identification of these subsystems. In this identification methodology, input/output data for each subsystem is necessary. The constrained least-squares stochastic subspace identification method (CLS-SSI) of Katayama and Picci (1999) is used as a tool for identification of the subsystems. An application to a river water quality modeling problem is proposed and results are presented.