摘要
为了在线快速评估当前电力系统稳定裕度,将回归分析和人工神经网络模型同时应用于电压稳定评估,用回归分析法来求负荷对电压稳定裕度的灵敏度,此种方法克服了传统潮流方法求取灵敏度的缺陷。为了改善神经网络模型的性能,根据预先设置好的灵敏度阀值来进行特征选择,从而减少输入变量的维数;然后设计了一个三层BP神经网络进行训练,求取电压稳定裕度,取得了很好的预测效果,最后在IEEE30节点系统上进行了验证。证明此方法的有效性。
In order to evaluate rapidly the voltage stability margin of power system,in this paper,a method of the regression analysis and neural network is proposed to estimate the voltage margin.The sensitivities are computed by using the regression models,which overcomes many limitations of the conventional methods of computing sensitivities.According to the sensitivities Threshold which is set in advance,the input variables is reduced largly.Three layers BP neural network was designed to calculate the voltage margin,which have a good performance of network.At last,the proposed method was test on IEEE30,to be proved the validity.
出处
《电力科学与工程》
2010年第8期19-23,共5页
Electric Power Science and Engineering
关键词
回归分析
灵敏度
特征选择
神经网络
稳定裕度
regression-analysis
sensitivity
feature selection
artificial neural networks
stability margin