基于曲线分段-聚类法的变电站并联补偿容量不等分组问题研究

摘要

For the problem of unequal grouping of parallel compensation capacity in substation, a new optimization method based on curve segmentation and clustering is proposed. Firstly, this paper calculates compensation capacity curve by transformer parameters and load curve and takes the maximal value as the sum of compensation capacity. Then it partitions this curve into several segments and clusters these segmentation results into K clusters, in which the number of cluster is equal to the number of capacitor groups, further modifies the clustering results in order to meet the condition of constraint, and takes the difference between two adjacent cluster results as the capacity of each group. In order to get the steady grouping results, it studies the relationship between segment number and grouping results. After the grouping plan is determined, taking the nine-area figure as control strategy. Finally, simulating with an equivalent practical power grid and load profile, the results show both the availability and rationality of this method that power loss is less than the method of equal capacity when capacitor is divided into 3 groups .%针对变电站并联补偿不等容优化分组问题，提出一种基于曲线分段-聚类法的优化分组算法。首先利用变压器参数与负荷曲线计算变电站总的无功需求曲线，将曲线最大值作为电容器总补偿量。然后利用曲线快速分段法和K-均值聚类法进行分段、聚类，使聚类数等于电容器分组数，并对聚类结果进行改进以满足各组容量之和等于总容量的约束，则各类间差值为分组容量。为得到稳定的聚类结果，以波动系数为指标研究分段数与分组容量结果之间的对应变化规律，给出选择分段个数的经验性结论。分组方案确定后，以九区图控制策略确定变压器分接头和电容器的投退。最后用算例验证了该方法的有效性及可行性，当电容器分组数相同时与等分组法结果比较，在动作次数相差不大的情况下，本方法可有效减小电能损耗。