基于灵敏度特征决策的GRNN短路容量智能辨识  

Intelligent Identification of GRNN Short-circuit Capacity Based on Decision-making of Sensitivity Characteristics

作　　者：潘睿[1] 刘俊勇[1] 郭晓鸣[1] 贺星棋[1] 刘友波[1] 方涛[1] 

机构地区：[1]四川大学电气信息学院,四川成都610065

出　　处：《四川电力技术》2010年第6期1-5,13,共6页Sichuan Electric Power Technology

基　　金：国家自然科学基金项目(No.50977059)

摘　　要：针对当前局部地区短路容量水平已接近现有设备额定值的情况,提出一种短路容量智能辨识方法。利用基于潮流的短路计算法计算系统各母线的最大短路容量,通过对典型潮流下灵敏度的计算,选择对短路容量贡献程度较大的发电机、负荷的有功出力作为输入特征向量,建立训练样本,对广义回归神经网络(GRNN)进行训练,构成该电网结构下的短路容量辨识的人工神经网络。应用该模型对运行中电网的母线短路容量水平进行快速扫描,为智能电网与智能调度中的故障识别快速仿真建模(FSM)提供了一种新思路。通过IEEE 30节点系统验证了该方法的可行性与有效性。Aiming at the situation that the level of short-circuit capacity is close to the rating of the existing equipment in local area at present,an intelligent identification approach of short-circuit capacity is proposed.Maximum short-circuit capacity of each bus is calculated using the short-circuit calculation method based on power flow.By calculating the sensitivity in typical power flow,the active efforts of generators and loads which have greater contribution to short-circuit capacity are selected as input eigenvectors,and the training samples are established to train general regression neural network（GRNN）.Then an artificial neutral network of short-circuit capacity identification is formed.The model is applied to fast scanning of short-circuit capacity of buses that are in operation in power grid,which is providing a new thought of fast simulation modeling（FSM） for fault recogintion in smart grid and intelligent dispatch.Simulations on IEEE 30 system are performed to verify the feasibility and validity of the approach.

关 键 词：短路容量 广义回归神经网络 灵敏度 智能辨识 快速仿真建模 

分 类 号：TM743[电气工程—电力系统及自动化]