基于改进k-means聚类的风电功率典型场景在日前调度中的应用

针对含风电电源的电网日前调度优化问题,应用聚类分析获得风电功率典型场景进行风电功率预测,并将预测结果用于日前调度优化具有重要意义。提出一种基于改进k-means聚类算法的风电功率典型场景生成方法,对周期内的风电数据通过场景生成和缩减,得到少数几个能反映周期内历史数据特征的风电功率典型场景集;然后以系统运行成本最小为目标,建立适应风电接入的日前机组组合模型,模拟风电接入后电力系统实际运行情况。最后通过算例比较风电功率点预测、区间预测和典型场景预测在电力系统日前调度中的经济运行优化结果,验证了所提方法的有效性和实用价值。

Application of Typical Wind Power Scenarios Based on Improved k-means Clustering in Day-ahead Dispatching

To solve the problem of day-ahead dispatching optimization of power grid containing wind power,it is of great significance to apply cluster analysis to obtain typical wind power scenarios for wind power prediction and apply the predicted results to day-ahead dispatching optimization.This paper proposes a generation method of typical wind power scenarios based on the improved k-means clustering algorithm.Through scene generation and reduction of wind power data within the cycle,a few typical wind power scene sets that can reflect the historical data characteristics within the cycle are obtained.Then,aiming at the minimum operating cost of the system,a day-ahead unit combination model suitable for wind power access is established to simulate the actual operation of the power system after wind power access.Finally,by comparing the economic operation optimization results of wind power point prediction,interval prediction and typical scenario prediction in day-ahead dispatching of power system,the effectiveness and practical value of the proposed method are verified.