Abstract: Wind power is characterized by intermittent, randomness, volatility and accurate prediction with little possibility. The large capacity wind power will bring serious challenges to the safety and stability of power system. According to the chaotic characteristic of wind speed, a new short-term prediction method of phase space reconstruction theory is put forward. The optimal delay time τ and embedding dimension m of the wind speed must be determined firstly. Then match m with τ, and find a best matching sample to reconstruct the phase space, finally using the BP neural network to forecast the short-term wind speed. By simulating the measured data of a wind power plant in Wulong region of Chongqing, the effectiveness and feasibility of the method is proved, and the accuracy of short-term power generation prediction is improved. It is also of great significance for the operation of grid connected wind power generation system.

Key words: Phase space reconstruction; complex autocorrelation method, false zero method, BP neural network, wind speed forecast

摘要： 风力发电具有随机性、间歇性、波动性和难以精确预测的特点，针对大容量风力发电接入,会对电力系统的安全、稳定运行带来严峻挑战的问题，根据风速的混沌特性,提出采用一种相空间重构理论短期预测方法。首先确定风速的最佳延迟时间τ和嵌入维数m，再次对m和τ的多组可行匹配,并找出一个最佳匹配进行样本相空间重构，最后使用BP神经网络进行短期风速预测。通过对重庆武隆地区某风电场的实测数据进行仿真，证明该方法的有效性和可行性，并且提高了短期发电功率预测精度，对并网风力发电系统的运行具有重要意义。

关键词: 相空间重构, 复自相关法, 虚假零点法, BP神经网络, 风速预测