Load Balancing In The Smart Grid With A Large Scale Penetration Of Electric Vehicles

With the improved human awareness of fossil shortage and severe environmental pollution, electric vehicles gain the attention of the government and the public because of its low carbon, energy saving, and environmental protection advantages. Electric vehicle is a kind of vehicle which uses battery-stored energy to support its daily trip-s, therefore, the emergence of electric vehicles can help reduce emissions of pollut-ing chemicals and reduce human dependence on fossil fuels. Our country supports the development of electric vehicles in many aspects, including developing production technologies, building charging stations and so on. However, it will create many new challenges for the operation and management of the power grid when a large scale of electric vehicles plug in the power grid, including increase peak load, increase energy loss and affect the energy quality.To deal with the created challenges and problems caused by the plugged electric vehicles, our work is listed in the following:1. The scheduling problem of electric vehicles charging is described as an optimal control problem in this paper, the objective of this problem is to flatten load dis-tribution in the power grid. What’s more, we explore the properties of the optimal charging method.2. We propose a totally distributed algorithm ODC to solve this problem iteratively. In each iteration, electric vehicles update their charging activities according to the control signal broadcasted by the power grid, the power system guide the activities of electric vehicles by altering the broadcasted control signal. No matter parameters of the electric vehicles are the same or not, this algorithm can finally converge to an optimal charge distribution solution. And finally we prove the optimality and convergence of algorithm ODC.3. Improve the above algorithm ODC and get a new algorithm AODC, this im-proved algorithm suits for asynchronous environment. Because it is possible that electric vehicles or the power system can not update their information due to the communication or some other reasons in some iterations in reality. Instead, they use outdated information to update their activities. The improved algorithm AODC can deal with this kind of situation.4. The electricity data in American PJM region is used as our base data, we compare the performance of algorithm ODC, AODC, MCH, and analyze them from the aspects of optimality and convergence.To our delight, our propose algorithm is totally distributed, each electric vehicle only needs to compute its own local problem, there is no interference between electric vehicles, so the communication and computation cost are very low. This kind of total distributed algorithm can easily extend to a large scale of problem which can support a large amount of electric vehicle emerged into the market. Theoretical and experi-mental results have shown that our distributed algorithm is effective to smoothen load fluctuation in the power system.