On Demand Response Algorithms And Test Platform For Smart Grid

Smart grid is an intelligent power grid that integrates advanced sensing, control, and communication technologies to enhance grid flexibility, efficiency, interactivity and cleanness, and it gets more and more attention from both academia and industry. Because smart grid is in its infancy, it has many challenges to be solved. Demand response which refers to schedule the power consumption manners of end-users to solve the contradiction between supply and demand is one of the key technologies in smart grid. In addition, as a bridge between theory study and practical application, the laboratory-based test platform plays a very important role in the research for smart grid. The research of the two aspects has been relatively mature in the traditional power grid, but the smart grid will fundamentally change the paradigm of the power grid from one-way energy and information flow to two-way energy and information flow, emphasizing the interations among end-users and energy utilities. The advantages of the smart grid technologies put forward new requirements in the demand response and the laboratory-based test platform. It is important to activate the participation of the end-users and special attention should be paid to protect the end-user privacy in the studying the demand response algorithms. Likewise, it needs to concern the integration between power grid and information network when developing the smart grid test platform. After summarizing the exsiting results, this thesis makes deep research on the two aspects according to the new requirements. The achievements are follows:1. With an emphasis on the nonshitable but power-adjustable electrical devices, a convex optimization problem which takes the social-welfare as an objective function is formulated, and then a newton-lagrange based distributed demand response algorithm (NDDR) is proposed. The global convergence of the NDDR algorithm is proved under milder conditions.2. For the unsatisfactory convergence rate of NDDR, the demand response for social-welfare optimization problem is divided into several sub-processes, and then a fast distributed demand response algorithm that achieves suboptimal social-welfare while ensures all the constraints is proposed through utilizing the gaussian belief propagation solver and distributed primal-dual interior method. The impact of the errors caused by the distributed computations is also analyzed theoretically.3. With an emphasis on delay tolerant electrical devices, the problem of scheduling the operations of these smart devices is formulated as a stochastic optimization model, and then a real-time distributed demand response algorithm for multi-users is proposed by utilizing Lyapunov optimization method. The proposed algorithm reduces the expected electricity cost with real-time electricity pricing and the influence on the users by having to give up part of the electricity request because of short supply of electricity or overload of the grid capacity. An explicit trade-off between the value of cost function and the worst-case delay of any electricity to be compledted or dropped is obtained through theoretical analysis.4. Adopting the idea of co-simulation, a smart grid common open research emulator (SCORE) which is built upon an open source communication network emulator is presented. SCORE emulates both information network and power grid and their integration. Further, based on the generalized branch cutting method, SCORE is designed to support large scale emulation and it can be run across multiple computers in the internet. SCORE has the key features of protability and scalability.5. According to the requirement of two-way energy and information flow in smart grid, a laboratory-based hardware testbed for smart grid is designed. The testbed consists of an information collection subsystem based on wireless sensor network and a power transmission/control subsystem. The information collection subsystem is used to monitor the environment, e.g. nobody is in the room with light on and the air-condtior is set too low, to help adjust operation patterns of the electricial devices. The power transmission/control subsystem contains a power network with energy flow and an information network that is responsible for sensing and controlling data flow.The obtained results are verified by simulations or experiments. Finally, the conclusion and future work are presented.