摘要
提出了基于蚁群优化算法和k阶近邻法相结合的嵌入式特征选择算法。选择稳态潮流量构成电力系统暂态稳定评估的输入特征集,针对输入特征集包含的大量冗余信息,特征选择结果中可能包含一定冗余特征的缺陷,先用聚类的方法裁剪冗余性特征,然后用所提算法选择和稳定状况强相关的关键特征,提高了特征选择的效率。通过对3机9节点和10机39节点系统进行计算,用所选择特征形成的规则能很高精度地判定系统的稳定水平,结果验证了所提方法的有效性。该算法可作为一种通用方法应用于不同规模的系统。
Embedded feature selection method based on ACO and k-NN is presented.Steady-state flow is selected as input feature set for transient stability assessment of power system.In view of large amount of redundant information in input feature set,there may be some redundant defects in selected features.The clustering method is firstly adopted to cut out redundancy and key features strongly correlated to stability are to select through mentioned method,which can enhance efficiency of feature selection.By calculation on 9-bus system of machine 3 and 39-bus system,the regulations formed by selected features can accurately judge stability level of the system and the result validates efficiency of the method.The method can be generally used in systems of different scales.
出处
《广东电力》
2011年第12期29-35,共7页
Guangdong Electric Power
基金
国家自然科学基金资助项目(50407014)
