基于交叠概率的含风电的地区电网暂态电压稳定评估的特征选择

Overlapped Probability Based Feature Selection for Assessment on Transient Voltage Stability of Regional Power System Containing Wind Farms

以选取风电地区暂态电压稳定特征量为目标,以最小落入交叠空间的概率为判据,从描述复杂电力系统的高维特征量中选取最简特征集,简化对风电地区暂态电压稳定的分析,同时达到了降低特征指标维度的目的。介绍了交叠概率算法的理论,运用交叠概率算法对特征量进行提取,对比标准39节点的电网和加入风电的标准39节点风电电网提取的交叠空间及选取的最简特征集,比较结果表明,用由交叠概率算法从高维特征量中选取的低维特征量组成的最简特征集描述同样的电力系统对象,也能具有较高的准确率。最后用神经网络验证选取的最简特征集的准确性。

Taking the characteristic quantity of transient voltage stability of regional power system containing wind farms as objective and taking the minimum probability of falling into overlapping space as the criterion, the simplest feature set is selected from high-dimensional characteristic quantities describing complex power grid to simplify the analysis of transient voltage stability in regional power system containing wind farms, in addition to achieve the aim of reducing the dimensionality of feature indices. The theory of overlapping probability algorithm is presented, and then characteristic quantities are extracted by overlapping probability algorithm. The overlapping space extracted from standard IEEE 30-bus system, the overlapping space extracted from IEEE 30-bus system with wind farm appended and the selected simplest feature set are compared, and the comparison results show that the description of the same power grid by the simplest feature set, which is composed by characteristic quantities with lower dimensionality selected from high-dimensional characteristic quantities by overlapping probability algorithm, is of higher accuracy. Finally, the accuracy of the selected simplest feature set is validated by neural network.