| The complexity of a complex system composed of a large number ofinterconnected dynamic elements may be significantly determined by the informationstructure constraints among the subsystems, as well as the information structuredynamic may emerge in abundant system properties. So it is of great significance toanalysis on the information flows. The main issues discussed in this dissertation arethe impact of information structure constraints variation on the interconnected systemsynthesis, as well as the control implementation of inclusion principle under thisdiscussion. The main contributions of this dissertation are stated as follows:(1) A group of structure transformation matrices are produced, in order to describethe information structure constraints dynamic of overlapping interconnected systems.The matrix description of information structure constraints dynamic in complexsystems is considered, basing on the permuted inclusion principle and its recurrentreverse order, so that this dynamic can be involved directly in the coordinated controldesign procedure. A pair of column and row group transformation matrices is definedto obtain the permutation matrix together with the expanded and contractedtransformation matrices applied in pair-wise decomposition after the system structuralvariation, by identifying the position data of system dynamic in permutation. In thisway, this dynamic can be described by the transformation matrices modification.(2) The locomotion consensus problem of a multi-agent system is considered underthe concept of system inclusion, as well as the circumstances that a new agent jointsthe overall system.The variation of vertexes in the system network topology may change theinformation structure constraints among them, and the most directly impact it bringsis the adjustment of information structure constraints within the neighborhood ofvaried vertexes. In this dissertation, a locomotion consensus problem of a multi-agentsystem is provided as an implementation example, with each pair of agentscoordinated concurrently by the pair-wise decomposition, and the structure addingmatrices can be constructed in order to deal with the circumstance that a new agent isaccreted to the overall system.(3) Based on the weighted Voronoi diagram, the potential division problem ofvertices in the system network topology is presented, moreover the influence of thevertices’ own properties on the system information structure constraints is alsoprovided.Each vertex in the system network topology may be presented in a particularpotential appearance. The properties of its own may also affect the informationstructure constraints of the overall system, on the division of influence area in thesystem network. By considering the network structure, as well as operating requirements of the smart grid, an optimal load configuration method based on theweighted Voronoi diagram is provided here. And this proposed method is mainlymotivated by the thought to divide the influence area of each substation within thepower grid. The capacity of each substation can be modified to adjust its potentialappearance, so that the overall system network topology is reconstructed in order toachieve better performance.(4) For the interconnected system with uncertain information structure constraints,the theoretical discussion on the implementation of pair-wise decomposition isprovided.The implementation of pair-wise decomposition is considered on an interconnectedsystem with uncertainties. Under the concept of system inclusion, consider twosystems with a same expanded system achieved by the same expand transformation asapproximations. It is proven that a coordinated controller can be found to stabilizeboth the two systems. This controller is contracted from the coordinated controller ofexpanded system, with each pair-wise subsystem sharing information structureconstraint taken into consideration.The work provided in this dissertation would be conducive to the development ofcorresponding theories. |
PDF Full Text Request