基于决策树的电力系统实时动态安全评估方法研究  

Research on real time dynamic security assessment method of power system based on decision tree

作　　者：蔡忠林 Cai Zhonglin(State Grid Baoji Electric Power Supply Company,Baoji 721004,China)

机构地区：[1]国家电网宝鸡供电公司,陕西宝鸡721004

出　　处：《能源与环保》2021年第5期202-207,共6页CHINA ENERGY AND ENVIRONMENTAL PROTECTION

基　　金：国家电网宝鸡供电公司项目(5226BJ1900BC)。

摘　　要：电力需求增长和电力市场管制的压力下,电力系统必须通过缩小运行安全裕度,以使其运行接近稳定极限,为系统实时安全指标提供充足的时间进行分析、决策和准确地实施补救控制,提出了一种对称不确定性(SU)算法和逻辑模型树(LMT)算法分别作为特征选择的高级分类器和决策树分类器,该方法利用对称不确定性(SU)来降低基于决策树分类器的动态安全评估(DSA)工具中的数据冗余。结果表明,SU显著降低了DSA数据集的维数,对于改进的IEEE 30总线测试系统模型的DSA,SU算法可以减少30.76%的计算时间,而LMT算法的精度可以提高到100%,同时提高了决策树分类器的性能。基于SU的决策树分类器能够近实时地评估系统的动态安全性。该方法对电力系统实时保护和控制应用具有一定的参考价值。Under the pressure of power demand growth and power market regulation,the power system must reduce the operation safety margin to make its operation close to the stability limit,and provide sufficient time for system real-time security index analysis,decision-making and accurate implementation of remedial control.A symmetric uncertainty(Su)algorithm and logic model tree(LMT)algorithm are proposed as feature selection this method uses symmetric uncertainty(Su)to reduce the data redundancy in the dynamic security assessment(DSA)tool based on decision tree classifier.The results show that Su algorithm can significantly reduce the dimension of DSA data set.For the DSA of the improved IEEE 30 bus test system model,Su algorithm can reduce the calculation time by 30.76%,while the accuracy of LMT algorithm can be improved to 100%,and the performance of decision tree classifier is improved.The decision tree classifier based on Su can evaluate the dynamic security of the system in near real time.This method has a certain reference value for the application of power system real-time protection and control.

关 键 词：动态安全评估 决策树 高级特征选择 对称不确定性 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TM33[自动化与计算机技术—控制科学与工程]