Decentralized stochastic adaptive control of dynamic non Gaussian systems

In this paper, the problem of stochastic adaptive control for a class of large-scale systems is addresed. As the locaal controllers is used a minimum variance controllers. The subsystems are described with single-input single-output ARMAX models. The stochastic disturbance is non-Gaussian and interconnections of subsystems are non linear functions. For estimation of unknown parameters of the local controllers projections stochastic approximation type algorithm is used. Nonlinear transformation of prediction error in algorithm is consequence of non-Gaussianity of disturbance. It is shown that the closed-loop system is globally stable and that the overall mean-square tracking error is bounded.