ABSTRACT
We initiate our guided tour through the development of control problems of large-scale systems, where we focus in Chapter 4 on conditions and algorithms for obtaining the dynamic controllers to solve the decentralized stabilization and the robust decentralized servomechanism problems (DSMP). The associated attributes including decentralized fixed modes of interconnected dynamical systems are carefully examined. We recall that in Chapter 3, the fundamental notion of decentralized control and its associated structures was briefly addressed. As it was introduced there and will be emphasized again here, the main motivation behind decentralized control is the limitations and/or failure of conventional methods of centralized control theory. Some basic techniques such as eigenvalue assignment, state feedback, optimal control, state estimation, and that similar to the latter (centralized control) theory demand complete information flow from all system sensors for the sake of feedback control. Clearly, these schemes are totally inadequate for feedback control of large-scale systems. Due to the physical configuration, high dimensionality, and/or interconnection patterns of such systems, a centralized control would be truly complex, which is neither economically feasible nor even necessary. Therefore, in many applications of feedback control theory to linear large-scale systems some degree of restriction is assumed to prevail on the information processing. In some cases a total decentralization is assumed; that is, every local control uj is obtained from the local output yj and possible external input wi [307, 337]. In others, an intermediate restriction on the information is possible. Some related issues and discussions were reported in [5, 43, 45, 106, 303, 335].
