Abstract. Renewable new energy power generation prediction error like wind and light, brings difficulties to dispatch the power system. In this paper, a multi-time scale robust scheduling method is set to solve this problem. It reduces the impact of clean energy prediction bias to the power grid by using multi-time scale (day-ahead, intraday, real time) and coordinating the dispatching power output of various power supplies such as hydropower, thermal power, wind power, gas power and. The method adopts the robust scheduling method to ensure the robustness of the scheduling scheme. By calculating the cost of the abandon wind and the load, it transforms the robustness into the risk cost and optimizes the optimal uncertainty set for the smallest integrative costs. The validity of the method is verified by simulation.
IntroductionWind power, photovoltaic, small hydropower and other renewable energy output power is susceptible to climate, altitude, topography, temperature and other natural factors, so has randomness and volatility. Currently under the condition of satisfying the safe operation, renewable energy adopts full internet access to stabilize its output fluctuations by thermal power, larger hydropower and other conventional power. However, the adjustment capacity of conventional power is limited. When the deviation between the output of renewable energy and forecast value is large, system active power is imbalance and trend is overload. And there will be abandon wind, abandon light, abandon water etc. In the field of operation control, it has been a point that how to improve renewable energy consumptive through the multi-time scale joint dispatch of renewable energy [1].In the paper [2], based on the difference in corresponding adjustment ability of power grid and wind power error at different time scales, a multi-time scale coordinated flexible load response scheduling model of "multi-level coordination, step-by-step refinement" is set to divide the whole process into dayahead plan (24h), intraday 4h rolling plan and real-time 15min plan and automatic power generation control. In the paper [3], wind power and load prediction value are expressed by fuzzy parameters. Based on the credibility theory, the traditional deterministic system constraint is transformed into fuzzy chance constraints, and dynamic economic dispatch model following that is established, which includes several fuzzy parameters at the multi-time scale. In the paper [4], the stochastic programming is used to reflect the probability of wind power. The unit commitment problem is described as a two-stage stochastic