基于门控递归单元神经网络的风速误差修正模型短期风电功率预测

Short-term Wind Power Prediction ofWind Speed Error Correction Model Based on Gated Recursion Unit Neural Network

随着风力发电的日益普及,风电功率预测已成为辅助电网调度和电力交易的基础。针对短期风电功率预测问题,提出一种基于门控递归单元神经网络的数值天气预报风速误差修正模型。首先,提取数值天气预报风速误差的标准差作为权重,并根据数值天气预报风速时间序列对这些权重进行重新排列,得到权重时间序列。然后,提出基于双向门控递归单元神经网络的误差修正模型,以数值天气预报的风速、趋势和权重时间序列为输入,对数值天气预报中的风速误差进行修正。利用修正后的风速,将风功率曲线模型应用于短期风功率预测。最后,利用风电场的实际数据与基准模型进行了有效性比较。结果表明,所提出的模型优于基准模型。

With the increasing popularity of wind power generation,wind power forecasting has become the basis of auxiliary power grid dispatching and power trading.In this paper,a wind speed error correction model of numerical weather forecast based on gating recursion unit neural network is proposed for short-term wind power prediction.Firstly,the standard deviation of the wind speed error of numerical weather forecast is extracted as the weight,and these weights are rearranged according to the wind speed time series of numerical weather forecast to obtain the weight time series.Then,an error correction model based on bi-directional gating recursion unit neural network is proposed,which takes the wind speed,trend and weight time series of numerical weather forecast as input to correct the wind speed error in numerical weather forecast.Using the modified wind speed,the wind power curve model is applied to short-term wind power prediction.Finally,the effectiveness of the proposed model is compared with the benchmark model by using the actual data of wind farm,and the results show that the proposed model is better than the benchmark model.