|Institution:||University of Minnesota|
|Full text PDF:||http://hdl.handle.net/11299/175387|
Optimal management plays an indispensable role in judiciously allocating the surging demand of limited resources available to our modern society. Intelligent management schemes must be efficient, scalable, and even robust to the inherently uncertain and possibly adversarial nature. Leveraging state-of-the-art optimization and signal processing techniques, the present thesis addresses several fundamental issues and emerging challenges of cyber-physical systems, especially for the smart power grid and wireless networks. Robust energy management is first dealt with for a grid-connected microgrid featuring distributed energy sources. To address the intrinsic challenge of maintaining the supply-demand balance due to stochastic availability of renewable energy sources (RES), a novel power scheduling strategy is introduced to minimize the microgrid operational cost including the worst-case energy transaction cost. The resulting optimization problem is solved in a distributed fashion by each local controller via the dual decomposition approach. In addition, for an islanded microgrid or the long-term planning of the bulk power system, risk-limiting energy management using the loss-of-load probability is developed. Day-ahead stochastic market clearing with high-penetration wind energy is further pursued based on the DC optimal power flow model. Capitalizing on the conditional value-at-risk, the novel model is able to mitigate the potentially high risk of the recourse actions to account for wind forecast errors. To cope with possibly large-scale dispatchable loads, fast distributed solvers are developed with guaranteed convergence. This thesis also caters to distributed resource allocation in wireless networks. Robust transceiver design and energy scheduling are considered for multiple-input multiple-output cognitive radio networks, as well as smart-grid powered coordinated multipoint systems. Robust optimization problems are formulated to tackle the uncertainties from imperfect channel state information and the nondispatchable RES. Efficient distributed solvers are tailored to the resulting convex programs through the techniques of semi-definite relaxation, primal, and dual decomposition. Numerical results are reported to corroborate the merits of the novel framework, and assess performance of the proposed approaches.