基于改进Shapley值的风电汇聚趋势性分状态量化方法

Research on Sub-state Quantization Method of Wind Convergence Trend Based on Improved Shapley Value

风电输出功率具有波动性,由于各机组出力之间的平抑效果,随着风电集群规模的增大,风电输出功率波动逐渐变缓,风电输出功率表现出"汇聚效应"。把握汇聚效应的趋势性对于规划送出线路及网架结构具有重要的意义。基于改进Shapley值法对风电汇聚效应的趋势性进行量化分析。在对不同风电输出状态量化分析的基础上,得到各状态下的持续出力曲线,进而构建基于汇聚特性分析的风电持续出力曲线分状态组合预测模型,并建立预测精度评价体系。采用改进的Shapley值法确定预测模型中的权重系数,避免了传统Shapley值法在单一模型预测结果偏差过大时仍参与组合的现象。基于实测数据对模型有效性进行检验,算例分析表明:相对于单一的预测模型,风电持续出力曲线的分状态组合预测方法能较更准确地描述风电汇聚的趋势。

Wind power output is fluctuant. Due to moderating effect of the output of each unit, the output fluctuation of wind power gradually slows down with wind power scale increase, and the wind power output shows a "convergence effect". It is of great guiding significance to grasp the trend of convergence effect for planning of outgoing transmission lines and capacity configuration. In this paper, the wind power output states are defined and the convergence trend of wind power in each output state is combined. The weight coefficients in the combination forecasting model are determined with the improved Shapley value method, avoiding the phenomenon that the traditional Shapley value method still participates in the combination when the deviation of a single model is too large. Based on quantitative calculation of continuous output curve of each state, a combined forecasting method of wind power continuous output curve is put forward based on analysis of convergence characteristics, and a prediction accuracy evaluation system is established. Validity of the method is verified with measured data. Case study shows that compared with the single prediction model, the combined forecasting method of wind power continuous output curve can accurately describe the trend of wind power convergence.