摘要
为了快速准确地定位电网强迫功率扰动源,根据"先感知,再分类,后定位"的思想,提出基于广域测量系统的空间特征椭球和决策树混合定位扰动源的新方法。通过对比分析不同强迫功率振荡信号,将实测的不同受扰轨迹信息映射到多维特征椭球,通过计算椭球的空间形状及其形态参数变化,实现强迫功率振荡态势的定量化描述;在抽取空间椭球特征参数的基础上,将不同扰动下强迫振荡瞬态阶段的特征椭球参数形成决策树样本集,利用C4.5算法离线训练,在线匹配以快速分类定位扰动源。算例结果表明,该方法可以在强迫功率振荡瞬态阶段快速分类定位不同扰动源,定位振荡主要参与机组和负荷的准确率很高。
According to the concept of ＂detection-classification-identification＂,a hybrid method applying space CELL（ Characteristic ELLipsoid） and decision trees based on the wide area measurement system is proposed to quickly and accurately identify the disturbance source of grid forced power oscillation. Different forced power oscillation signals are compared and the information of different measured perturbed trajectories is mapped to a multi-dimensional CELL. The space of CELL and its shape parameter variation are calculated to realize the quantitative description of forced power oscillation. The characteristic parameters of CELL during the transient phase of forced power oscillation under different disturbances are extracted to form the decision tree set,which is offline trained by C4.5 algorithm and online matched to fast classify and locate the disturbance source. Results of case study show that,the different disturbance sources are quickly classified and identified during the transient phase of forced power oscillation and the main contributing units and loads are identified with very high accuracy.
出处
《电力自动化设备》
EI
CSCD
北大核心
2015年第2期125-132,共8页
Electric Power Automation Equipment
基金
国家自然科学基金资助项目(51261130472)
英国ALSTOM公司国际科研合作项目(11H0207)~~
关键词
强迫振荡
扰动源定位
广域测量系统
特征椭球
决策树
forced oscillation
disturbance source identification
wide area measurement system
characteristic ellipsoid
decision tree
