基于时间序列模型的风速预测方法VIP专享

I摘 要要建立风电场需具有 30 年以上的测风数据，且连续性要比较好，要建立这样一个风电场需要很长时间，用时间序列法预测风速能有效减少测风时间。本文利用差分自回归滑动平均模型对风速进行预测，在已知历史风速的基础上对未来风速进行预测。在 MATLAB 平台先对瓜州气象站 1978~2007 年的平均风速进行预测，然后对瓜州气象站某一年前 144 天的日平均风速进行预测，并对其预测结果进行分析。由残差图来看此模型拟合精度较高，拟合效果较为理想。关键字：差分自回归滑动平均模型；风速；MATLABAbstractTo build a wind farm requires more than 30 years of wind measurement data, and the continuity is better. It takes a long time to build such a wind farm. Using the time series method to predict wind speed can effectively reduce the wind measurement time. In this paper, the differential autoregressive moving average model is used to predict the wind speed, and the future wind speed is predicted based on the known historical wind speed. The MATLAB platform first predicts the average wind speed of Guazhou Meteorological Station from 1978 to 2007, and then predicts the average daily wind speed of the Guazhou Meteorological Station for 144 days a year ago, and analyzes its prediction results. Judging from the residual plot, the fitting accuracy of this model is high, and the fitting effect is ideal.Keywords: differential autoregressive moving average model; wind speed; MATLAB§ 1 绪 论1.1 课题所研究的意义和目的风能作为可再生能源和清洁能源且资源丰富，因而风电在世界各地迅速发展。但风力发电也存在着明显的缺点，它的大问题就是风机出力的不稳定性和波动性。风速直接影响风机输出的功率，在风力发电中。风机输出的电功率具有随机性，因为风速含有随机波动性，且无法准确的预测，这种随机性对含有风电并网的电力系统以及风电机组的运行和控制会产生不利影响。而且在建一个风电场需持续测风十年以上，若达到条件才可以建风电场，周期太长，太耗费人力物力。而预测风速便可以很有效的解决这一问题，已...