Decentralized control of large-scale systems based on canonical representation

A stable decentralized control scheme for a linear time-invariant discrete-time system with zero-mean random disturbance is proposed. The proposed scheme is based on similarity transformation of a one-way-observable canonical form. Local control sequences of each station are generated by exploiting the structure of the first subsystem of the transformed model and this, in turn, is used to compute the transformed-model local controller and estimator gains in a decoupled manner. The proposed scheme reduces the overall computational burden and telemetry costs, as compared with those of the centralized LQG control scheme. A suitable example is considered to test the proposed scheme and the results of the simulation are presented.