关键词: 虚拟电厂, 经济模型预测控制, 非合作博弈, 自适应步长, 优化调度

Abstract:

Considering the uncertainty of new energy's output, economic and optimal operation of multiple virtual power plants have attracted considerable attention. Herein, a two-stage coordinated optimal scheduling method for multiple virtual power plants based on economic model predictive control is proposed. First, distributed power models, including wind turbines, gas turbines, and energy storage batteries models, are established, and an interruptible load model that participates in demand response is developed. Then, the day-ahead scheduling maximizes the revenue of multiple virtual power plants, employing Nash optimal decision-making to achieve optimal economic scheduling of multiple virtual power plants based on improved economic model predictive control. The length of the prediction domain is adaptively selected as per the wind power forecast error to reduce the impact of wind-power forecast uncertainty on the dispatch model. Furthermore, a real-time feedback scheduling strategy considering the economy is developed to make up for the power deviation of the day-ahead. Analysis of the case demonstrates that, compared with a single virtual power plant trading with only a large power grid, the total profit of the joint operation of multiple virtual power plants increases by 26.75%. Furthermore, compared with the fixed step size, the transaction power error is reduced by 78.28% under the adaptive domain step size. The proposed scheduling scheme effectively copes with the uncertainty of wind power and improves scheduling accuracy while realizing the overall optimal economic operation of multiple virtual power plants; thus, the correctness and feasibility of the scheduling model are confirmed.

Key words: virtual power plant, economic model predictive control, non-cooperative game, adaptive step size, optimal scheduling