Model And Algorithm On Unit Commitment Considering The Influence Of Wind Power Forecast Error

Energy is the key of social and economic secure sustained and fast growth, but the traditional energy supply system is facing serious challenge for the growing fossil energy shortage and worsening environmental pollution. So the clean energy like wind power, solar power and so on has already caused wide concern. Wind power has been one of the most important energy of power system because of its mature technique and economical efficiency. However the wind power brings great challenges for the dispatching and operation of power system on account of the uncertainty of wind, which reduces carbon emissions and supply clean energy.On the one hand, it is hard to grasp the randomness of wind power. Although the forecast technique of wind power has made great development in this years, it is still difficult to meet the actual engineering requirement. So it is important to reduce impacts on unit commitment(UC) and economic dispatch by grasping the characteristics of wind power forecast error, on the basic of existing forecasting techniques. Obviously it is of significance for the power system safe and economic operation.On the other hand, the random wind power fluctuation is difficult to coordinate with the existing deterministic dispatching. For the growing wind power, the uncertainty of power system is increasing, even threating the power system safe operation, with great economical and environmental benefits. The traditional unit commitment and economic dispatch is not suitable for the modern power system with large scale wind power. It is of theoretical and practical significance to study unit commitment with large scale wind power for improving power system safety and economic.Therefore, this paper studies the characteristics of wind power forecast error, and analyzes how to use the error information into unit commitment problem. This paper proposes a unit commitment method considering the sequential characteristic of forecast error to improve the power system safety and economic. The main work can be summed up as follows:(1) To analyze the probability density function(PDF) of wind power forecast error, the parameter and non-parameter methods are used to fit the forecast error, like normal distribution, beta distribution, T-location-scale distribution and non-parameter kernel density estimation. An improved index is proposed to evaluate the fitting effort of the pdf, and compare the fitting effort of different fitting methods using actual forecast error data.(2) Considering the difference in power dimension and sequential dimension respectively, a novel segmentation fitting method is proposed, then consider the characteristics in power-sequential dimension at the same time. To improve the fitting effects and reduce the computation caused by the segmentation method, a bin reduction method is necessary, which makes the proposed method more practical. The analysis example with actual error data indicates that the proposed fitting method can describe the forecast error characteristics more accurately.(3) A novel unit commitment model considering forecast error sequential characteristic with reserve classification is proposed and solved. This model combines the forecast error sequential characteristics with UC model sequential characteristics, so the error information is accurately utilized by the UC model. Considering the pdf of forecast error, the uncertainty causing by wind is classified by confidence, coping with various reserve, generating corresponding costs, and the confidence intervals are automatically set by balancing these costs with traditional operation costs. The hybrid particle swarm optimization(PSO) algorithm with heuristic search method is used to solve this model, which verifies that the proposed model is better than before.