文章摘要
胡志鹏,刘剑,张玻,等.基于风险关键特征量的输电线路运行环境风险评估[J].电力系统自动化,2017,41(18):160-166. DOI: 10.7500/AEPS20161213002.
HU Zhipeng,LIU Jian,ZHANG Bo, et al.Risk Assessment of Operation Environment for Transmission Lines Based on Risk Key Characteristic[J].Automation of Electric Power Systems,2017,41(18):160-166. DOI: 10.7500/AEPS20161213002.
基于风险关键特征量的输电线路运行环境风险评估
Risk Assessment of Operation Environment for Transmission Lines Based on Risk Key Characteristic
DOI:10.7500/AEPS20161213002
关键词: 架空输电线路  风险评估  最小二乘支持向量机  主成分分析  风险关键特征量  降维  风险状态预测
KeyWords: overhead transmission line  risk assessment  least squares support vector machines(LS-SVM)  principal component analysis(PCA)  risk key characteristic  dimensionality reduction  risk state forecast
上网日期:2017-05-16
基金项目:国家自然科学基金青年基金资助项目(51007066)
作者单位E-mail
胡志鹏 武汉大学电气工程学院, 湖北省武汉市 430072  
刘剑 武汉大学电气工程学院, 湖北省武汉市 430072 lj_eec@whu.edu.cn 
张玻 国网四川省电力公司德阳供电公司, 四川省德阳市 618000  
王美荣 国网四川省电力公司德阳供电公司, 四川省德阳市 618000  
邝石 国网河南省电力公司郑州供电公司, 河南省郑州市 450000  
杨雪瑞 国家电网东北电力调控分中心, 辽宁省沈阳市 110180  
摘要:
      目前架空输电线路运行环境的整体风险评估主要基于复杂的危险源辨别和风险评估(HIRA)和累积扣分法,为了简化输电线路运行环境风险评估方法,建立了结合最小二乘支持向量机(LS-SVM)和主成分分析(PCA)的架空输电线路运行环境整体风险评估模型。首先,通过对故障跳闸情况进行统计分析,得到10种危害架空输电线路安全的运行环境技术因素,并提取出相应的表征风险特征量;考虑实际数据采集的工作量和风险评估的复杂性,使用PCA对特征量进行降维处理,得到权重较大的关键特征量;将这些风险关键特征量作为LS-SVM的输入,建立了风险评估预测模型。最后,以传统评估结果为样本给出评估模型训练和测试结果,结果表明模型预测分类的精度可以达到95%。
Abstract:
      Currently the risk assessment of operation environment for overhead transmission lines mainly based on complicated hazard identifying and risk assessment(HIRA)and the cumulative points methods. To simplify the histogram risk assessment of overhead transmission lines, a histogram risk assessment model of overhead transmission lines based on least squares support vector machines(LS-SVM)and principal component analysis(PCA)is developed. Firstly, through a statistical analysis of fault tripping, 10 kinds of running environmental technology factors that endanger the safety of the overhead transmission and corresponding characteristics are obtained. Considering the complexity and workload of the actual data collection and risk assessment, PCA is used to reduce the dimension of factors to obtain characteristics factors. These are used as inputs of LS-SVM to build a risk assessment forecast model. Finally, an example based on a traditional assessment sample is given to show the accuracy of results of the training and assessment model can reach 95%.
查看全文(Free!)   查看附录   查看/发表评论  下载PDF阅读器