| Wind power as the renewable energy with most mature technology and the most promising future,has made rapid development in the last 10 years.The proportion of wind energy accounted for energy consumption in China increased year by year and wind power is gradually becoming the major alternative energy.However,as a primary energy,wind power is uncontrollable and its power output is random and intermittent,which has brought great challenges to the traditional power grid scheduling decision-making.In this paper,research of the modeling and optimization algorithm of rolling and real-time optimal scheduling of power system scheduling with large-scale wind power incorporation has been carried out,the main contents and achievements of this paper are summarized as follows:Owing to the accuracy of point prediction of wind power can not meet the requirements of actual scheduling,it is necessary to consider the uncertain factors of wind power in scheduling decision.We need to find a suitable uncertainty expression and embed it in the scheduling model to giving dispatch the ability of dealing with uncertainty factors,that is,improve the scheduling model from the objective function dimension.In this paper,based on analysis of wind power forecast value and actual value,firstly we use the normal distribution function to describe the distribution of ultrashort prediction error of wind power.Then,by using the latin hypercube sampling and the synchronous iteration scene reduction technique,several wind power error scenarios are obtained and the probability distribution of continuous random variables are discretized.Then,the optimal selection method of units is proposed considering the wind power forecast error and the unit capacity and the power of the cross sections.A stochastic optimization expectation model is established considering the cost of regulation,coal cost and cost of wind abandonment,in which the wind power output is included in the decision variable and the planned value of wind power is given.The model takes into some complex constraints into accounted,such as the climbing ability limit of units,the tie line power limit and the planned curve ladder,etc.Finally,the paper proposes a modified particle swarm optimization algorithm to solve the problem.Power system active power optimization scheduling is a complex dynamic process of multi-level,multi-time scale of,and needs mutual coordination and cooperation of multiple time-level optimization and control.So it is neceseery to increase the days of ultra-short-term scheduling model to build multi-time scale scheduling model,which is extending the model from the time dimension.This paper compares the day ahead plan,the rolling/real-time plan and the AGC from different perspectives,and analyzes its internal relationl.Then,the coordination method of the upper and lower plan and AGC is proposed and on this basis the rolling optimal scheduling and real-time optimal scheduling are established as a transition between day ahead scheduling and AGC to eliminate the impact of wind power prediction error step by step.The actual example shows that the stochastic optimization model considers different wind power error scenarios,obtains the expected minimum value in each scene,the wind power optimization result and the short-term forecast value are deviated.The wind power optimization result is more consistent with the actual situation,and promotes wind power consumption and reduces the occurrence of wind power curtailment at the rolling plan level.The rolling and real-time scheduling considering the unit optimization avoids the large-scale adjustment of the power generation plan of the unit,realizes the automatic tracking and adjustment of the wind power forecast deviation on the ultra short-term time scale,improves the accuracy of the power generation plan and reduces the comprehensive power generation cost and the risk of cross-section power limit,improves the safety and economy of power gird. |
PDF Full Text Request