Research Of Optimization Algorithm To Load Distribution Of Peak Shaving Operation Among Multiple Provincial Power Grids

Peak load regulation is a common problem for many power grids in our country. With a rapid growth of electricity requirement from industry and civil usage, the power grid is facing a huge peak-valley difference. In addition, the increasing capacity rate of new energy which has the characteristic of reverse peak shaving is making the problem more severe. As a result, it plays a significant role in alleviating the increasing peak demands of electrical power systems to coordinate generation from high-quality power sources among provinces via the bulk power grid platform. How to distribute the generation among different province power grids determines the effective of peak shaving operation. However, the existing coordinating way and optimization method usually fails to produce satisfactory generation schedules which cause fluctuating loads over the whole operational horizon.In this paper, a method for formulating the peak shaving operation of power plants is proposed by introducing the ideal remaining loads (ideal solution) of each power grid. Ideal solution is usually more smoothed and with less peak-valley difference. As a result, minimizing the deviation between the current solution and the ideal solution is chosen as the objective function so that the remaining loads over local period of one day can be smoothed. In order to construct the ideal solution, this method homogenizes multiple remaining load values over contiguous periods by using two different ways, named N-neighbor combination and local combination. The N-neighbor combination determines the ideal value of one point as the average value of its N neighboring points. The local combination first splits the loads of each power grid into several sections and the average value of the points in the same section is the ideal value. The variable step neighborhood search algorithm is used to optimize the previous model. It’s an iterative algorithm and during the search process, the search step is dynamically updated to improve the convergence speed and solution quality. The model is implemented on the operation of power plants directly controlled by the East China Grid. The comparison with the existing method demonstrates the advantages of our method, and the results show that our method is capable of coordinating the different peak load demands among Shanghai, Jiangsu, Zhejiang and Anhui power grids.