贝叶斯网络及时间序列下电力系统元件故障诊断模型被引量：12

Fault diagnosis model for power system components based on Bayesian network and time series

下载PDF

导出

摘要针对电力系统频繁出现故障,其人工成本及工作危险系数增加的问题,在贝叶斯网络及时间序列下提出一种电力系统元件故障诊断技术.将提取的故障特征作为待诊断的特征项,确定每个类别下故障因子的集合,通过集合中潜在故障因子的系数判断真假,为真则将该类别下的因子系数代入到挖掘公式中,利用时间序列匹配法求得故障因子发生与未发生概率,根据单位元件发出的具体断开或跳闸等保护动作判断具体故障元件.仿真结果表明,所提出的故障诊断模型诊断结果与实际结果的拟合程度较高,拥有较高的准确率和有效性,在智能电网领域具有很好的应用潜力. Aiming at the problem of frequent faults in power system and the increase of labor cost and work risk coefficient,a fault diagnosis technology of power system components based on Bayesian network and time series was proposed.The extracted fault features were regarded as the feature items to be diagnosed,and the set of fault factors under each category was determined.The coefficient of potential fault factors in the set was used to judge whether it was true or not.If it was true,the factor coefficient under this category was substituted into the mining formula,and the occurrence and nonoccurrence probabilities of fault factors were obtained by using the time series matching method.According to the protection operation of unit component,such as breaking or tripping,the specific fault component can be judged.The simulation results show that the as-proposed fault diagnosis model has relatively higher fitting degree with the actual results and comparatively higher accuracy and effectiveness,and exhibits good application potential in the field of smart grids.

作者裴求根黄达文高祥斌 PEI Qiu-gen;HUANG Da-wen;GAO Xiang-bin(School of Electrical and Electronic Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Information Center,Guangdong Power Grid Co.Ltd.,Guangzhou 510600,China;Information Center of Zhaoqing Power Supply Bureau,Guangdong Power Grid Co.Ltd.,Guangzhou 510600,China;Computer Teaching and Research Office,Linyi University,Linyi 273400,China)

机构地区华中科技大学电气与电子工程学院广东电网有限责任公司信息中心广东电网有限责任公司肇庆供电局信息中心临沂大学计算机教研室

出处《沈阳工业大学学报》 CAS 北大核心 2021年第6期619-623,共5页 Journal of Shenyang University of Technology

基金山东省软科学课题(2018RKB01381).

关键词智能电网贝叶斯网络低频带时间序列匹配法潜在故障因子保护动作因子系数故障元件 smart grid Bayesian network low frequency band time series matching method potential fault factor protection action factor coefficient fault component

分类号 TM324 [电气工程—电机]