为了全面准确地检测出潜在的设备异常
针对汽轮机、风电转子等高关联耦合分布式系统多测点多源传感器产生的多维多态监测数据
提出了一种基于多变量耦合网络与变分图自编码器的异常检测方法。首先采用去趋势互相关分析(DCCA)
定量分析多维变量间的耦合关系
构建复杂系统多变量耦合关系网络。在此基础上
建立基于无监督学习的变分图自编码模型
对系统多变量耦合关系网络进行特征提取
使用正常数据训练该模型
图卷积网络作为编码器学习输入数据的分布
采样获得其潜在表示
以实现耦合网络的重构
采用重建概率作为系统多维多态监测数据异常检测评价指标。最后以某火电厂汽轮机组转子系统监测数据为例开展异常检测分析
结果表明:考虑多维多态监测数据间的耦合关系
提高了系统异常检测的准确性和可靠性;引入基于变分图自编码器的无监督学习方法
降低了经验依赖性并克服了异常样本少的问题。
赵晓军.时间序列的相关性及复杂性研究[D].北京交通大学,2015(06).
李杰,倪军,王安正.从大数据到智能制造[M].上海交通大学出版社,2016.
曹鹏.旋转机械状态监测方法研究及系统实现[D].西南石油大学,2014(03).
冯龙飞,高建民,高智勇,谢军太,高旭.基于DCCA-NSEn的系统耦合网络建模与评估[J].振动.测试与诊断,2019(05).
佘博,田福庆,梁伟阁.基于深度卷积变分自编码网络的故障诊断方法[J].仪器仪表学报,2018(10).
赵洪山,刘辉海.基于深度学习网络的风电机组主轴承故障检测[J].太阳能学报,2018(03).
刘国斌.多传感器信息融合与神经网络的汽轮机的故障分析[J].内燃机与配件,2017(22).
WANG RongXi,GAO JianMin,GAO ZhiYong,GAO Xu,JIANG HongQuan.Complex network theory-based condition recognition of electromechanical system in process industry[J].Science China(Technological Sciences),2016(04).
杜海峰,王娜,张进华,邵颉,王孙安.基于复杂网络的故障诊断策略[J].机械工程学报,2010(03).
姜洪权,高建民,陈富民,高智勇.基于网络特性的分布式复杂机电系统脆弱性分析[J].计算机集成制造系统,2009(04).
秦锋,杨波,程泽凯.分类器性能评价标准研究[J].计算机技术与发展,2006(10).
Niklas Renström,Pramod Bangalore,Edmund Highcock.System-wide anomaly detection in wind turbines using deep autoencoders[J].Renewable Energy,2020.
Department of Electrical and Electronic Engineering, North China Electric Power University, 071003, Baoding, China,Bohai New Area Power Supply Branch, Cangzhou City, State Grid Hebei Electric Power Co., Ltd., Cangzhou 061113, China,Department of Electrical and Electronic Engineering, North China Electric Power University, 071003, Baoding, China,Department of Electrical and Electronic Engineering, North China Electric Power University, 071003, Baoding, China.Anomaly detection and fault analysis of wind turbine components based on deep learning network[J].Renewable Energy,2018.
Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China,Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China,Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Tsinghua University, Beijing 100084, China,Department of Energy and Power Engineering, Tsinghua University, Beijing 100084, China,State Key Laboratory of Control and Simulation of Power System and Generation Equipment, Tsinghua University, Beijing 100084, China.An unsupervised spatiotemporal graphical modeling approach for wind turbine condition monitoring[J].Renewable Energy,2018.
Diederik P. Kingma,Tim Salimans,Max Welling.Variational Dropout and the Local Reparameterization Trick.[J].CoRR,2015.
Podobnik Boris,Stanley H Eugene.Detrended cross-correlation analysis: a new method for analyzing two nonstationary time series..Physical review letters,2008.
0
浏览量
3
下载量
3
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621