神经网络在变压器油中溶解气体微机在线监测中的应用研究
|
摘要
本论文研究应用变压器油中溶解气体分析方法诊断变压器故障，针对气体传感
器之间存在的“交叉敏感”特性提出了一种基于人工神经网络的智能嗅敏系统。研
究了采用各气敏传感器组成的气敏元件阵列与人工神经网络来实现混合气体的组分
与浓度识别的原理、方法和构成，提出了一种改进型误差反向传播网络训练方法，
它具有良好的泛化能力。由于B-P网络具有极强的学习能力，能够充分逼近任意复
杂的非线性关系，所以网络的学习过程是将隐含在训练样本中的气体之间复杂的非
线性关系分布到各连接权上，形成所谓的“知识”或“经验”，这些“知识”或“经
验”能较精确地反映复杂的气体信号，从而能构成对气体的较准确的分析测试。仿
真试验显示了这一方法的可行性与优越性。结合变压器油中溶解气体微机在线监测
项目讨论了系统硬件以及软件的具体实现。混合气体识别的传统方法——色谱法，
采用将各混合气体逐次分离后再进行识别，此法装置复杂、费用昂贵；多变量分析
模式识别法需要建立大量的气体响应方程，工作量非常大，使其实用范围受到影响。
采用神经网络进行识别则克服了上述种种不便，而且模拟运行结果表明该法识别速
度快、识别率较高，因其不再过分依赖传感器的选择性，是实现智能化检测混合气
体一种很有前途的方法。
In this paper the dissolved gas analysis is applied to diagnose
the transformer fault,
and an intelligent olfaction system based on artificial neural
network is also presented to
avoid the cross sensitivity among gas sensors. The principle,
method, and constitution of
mixed gases identification of elements and concentration are
studied by using gas sensor
array and artificial neural network. The improved error back
propagation that has good
generalization ability is adopted to train the neural network. As
B-P network has the
ability of study, it can approximate any complex non-linear
relation. The knowledge and
experience which is distributed in the weights among the neural
nodes is obtained by
training the neural network, which can reflect the non-linear
relation among the gas
signals and analyze the mixed gases more accurately. Simulation
experiment shows the
feasibility and advantage of this method. The hardware and
software of microcomputer
on-line detection of dissolved gas in transformer oil is also
discussed how to realize.
Chromatogram, the traditional, complex, expensive method,
identifies the mixed gases by
separating them. Multi-variable analysis pattern recognition is
narrowly employed because
abundant gas response equations are needed. The method of
artificial neural network not
only overcomes the above shortcoming but also identifies rapidly
and accurately.
Intelligent instrument of detecting mixed gases has a great
future because the selectivity of
sensors is not so necessary.
器之间存在的“交叉敏感”特性提出了一种基于人工神经网络的智能嗅敏系统。研
究了采用各气敏传感器组成的气敏元件阵列与人工神经网络来实现混合气体的组分
与浓度识别的原理、方法和构成，提出了一种改进型误差反向传播网络训练方法，
它具有良好的泛化能力。由于B-P网络具有极强的学习能力，能够充分逼近任意复
杂的非线性关系，所以网络的学习过程是将隐含在训练样本中的气体之间复杂的非
线性关系分布到各连接权上，形成所谓的“知识”或“经验”，这些“知识”或“经
验”能较精确地反映复杂的气体信号，从而能构成对气体的较准确的分析测试。仿
真试验显示了这一方法的可行性与优越性。结合变压器油中溶解气体微机在线监测
项目讨论了系统硬件以及软件的具体实现。混合气体识别的传统方法——色谱法，
采用将各混合气体逐次分离后再进行识别，此法装置复杂、费用昂贵；多变量分析
模式识别法需要建立大量的气体响应方程，工作量非常大，使其实用范围受到影响。
采用神经网络进行识别则克服了上述种种不便，而且模拟运行结果表明该法识别速
度快、识别率较高，因其不再过分依赖传感器的选择性，是实现智能化检测混合气
体一种很有前途的方法。
In this paper the dissolved gas analysis is applied to diagnose
the transformer fault,
and an intelligent olfaction system based on artificial neural
network is also presented to
avoid the cross sensitivity among gas sensors. The principle,
method, and constitution of
mixed gases identification of elements and concentration are
studied by using gas sensor
array and artificial neural network. The improved error back
propagation that has good
generalization ability is adopted to train the neural network. As
B-P network has the
ability of study, it can approximate any complex non-linear
relation. The knowledge and
experience which is distributed in the weights among the neural
nodes is obtained by
training the neural network, which can reflect the non-linear
relation among the gas
signals and analyze the mixed gases more accurately. Simulation
experiment shows the
feasibility and advantage of this method. The hardware and
software of microcomputer
on-line detection of dissolved gas in transformer oil is also
discussed how to realize.
Chromatogram, the traditional, complex, expensive method,
identifies the mixed gases by
separating them. Multi-variable analysis pattern recognition is
narrowly employed because
abundant gas response equations are needed. The method of
artificial neural network not
only overcomes the above shortcoming but also identifies rapidly
and accurately.
Intelligent instrument of detecting mixed gases has a great
future because the selectivity of
sensors is not so necessary.
引文
[1] 和田 昱二等编著，张家元，陈幼祥，陈进跃译．电气设备诊断技术及其自动化．北京：机 械工业出版社．1990,7．
[2] 潘笑，汪小澄．变压器绝缘油击穿电压自动测试仪．变压器.1998，35(4) ：32-34．
[3] 黎洪昌．35-110KV油浸式电流互感器介损测量时应注意的问题．变压器.1999，36(10) : 23-26．
[4] 董旭柱，王昌长，朱德恒．电力变压器局部放电在线监测研究的现状和趋势．变压器．1996，33 (5) ：2-7．
[5] 操敦奎．变压器油中气体分析与诊断，武汉：湖北新生报出版社．1987．
[6] SD187-86． 变压器油中溶解气体分析及判断导则．1986．
[7] 王宗．一种胶囊式储油柜的假油位分析．变压器．1996，33(3) ：39．
[8] 邬家义．国产变压器油发展概述．变压器1996，33(1) ：10-14．
[9] 陆万烈，夏业勤．变压器绕组温度测量的“热模拟”误差．变压器1999，36(10) ：15-17．
[10] 袁道君．油泵故障引起的变压器油金属污染及现场再处理．变压器1999，36(1) ：35-38．
[11] 严璋.电气绝缘在线检测技术．北京：水利电力出版社．1995．
[12] Tsukioka H．New Apparatus for Detecting Transformer Faults．IEEE Trans EI．1986，21(2) ，221．
[13] 李义仓．变压器油中CO和C02与绝缘利料劣化的关系及故障诊断．华东电力1993，21(2) ： 16-19．
[14] 张冠军，钱政，严璋．DGA技术在电力变压器绝缘故障诊断中的应用与进展．变压器1999，36 (1) ：30-34．
[15] 石伟峰．变压器油中气体的TD图分析法．华东电力. 1989，17(1) ：14-18．
[16] 高宁，高文胜，严璋．人工神经网络在电力设备绝缘诊断中应用的进展．高压电器.1997，33 (6) ：31-35．
[17] Lin C E, Ling J M, Huang C L. An Expert System in Transformer Fault Diagnosis Using Dissolved Gas Analysis. IEEE Trans. On Power Deliv. , 1993, 8 (1) .
[18] Tomsovic K, Tapper M, Ingvarsson T. A Fuzzy Information Approach to Integrating Different Transformer Diagnosis Methods. IEEE Trans. On Power Deliv., 1993, 8 (3) ,1638-1646．
[19] Chang C S. Thia B S. Online Rescheduling of Mass Rapid Transit System: Fuzzy Expert System Approach.IEE Proc. Electr. Power Appl. , 1996, 143 (4) , 307-316．
[20] 蔡超豪．利用人工神经网络对变压器故障进行早期诊断．变压器1997，34(1) ：34-37．
[21] 高宁，高文胜.严璋．基于模糊理论与自适应共振网络的油中气体分析诊断．高电压技术 1997，23(4) ：22-25．
[22] Penman J, Yin C M. Feasibility of Using Unsupcrviscd Learning, Artificial Neural Networks for the Condition Monitoring of Electrical Machines. 1EE Proc. Electr. Power Appl , 1994, 141 (6) ,317-322．
[23] McCulloch W S, Pitts W. A Logic Calculus of the Ideal Imminent in Nervous Activity. Bulletin of ath. Biophys. 1943,5:115-133．
[24] Hebb D D. The Organization of Behavior. New York: John Wiley & Sons, 1949．
[25] Rosenblatt F. Principles of Neurodynamics. New York: Spartan Book,1959．
[26] Widrow B, Hoff M E. Adaptive switching circuits. 1960 IRE WESCON Conv. Record. Part 4． 1960,8,96-104．
[27] 王永骥，涂健．神经元网络控制．北京：机械工业出版社．1998，2．
[28] 李孝安，张晓缋．神经网络与神经计算机导论．西安：西北工业大学出版社．1994，10．
[29] Zaromb S, Stetter J. Theoretical basis for identification and measurement of air contaminants using an array of sensors having partly over lapping selectivities. Sensors and Actuators. 1984,6,225-243．
[30] Garey W, et al. Chemometric analysis of multisensor arrays. Sensors and Actuators. 1986,9,223-234．
[31] Muller R, Lange E. Multidimensional sensor for gas analysis. Sensors and Actuators. 1986,9,39-48．
[32] Gall M, Muller R. Investigation of gas mixtures with different MOS gsa sensors with regard to pattern recognition. Sensor and Actuators. 1989,17,583-586．
[33] Hierold C, Muller R. Quantitative analysis of gas mixtures with non-selective gas sensors. Sensors and Actuators. 1989,17,587-592．
[34] Hornor G, Hierold C. Gas analysis by partial modclbuilding. Sensors and Actuators. 1990, B2, 173-184．
[35] R.Hecht-Nielsen. Theory of the Back Propagation Neural Network. Proc of IJCNN, 1989． 1．
[36] Anders Krogh, John A.Hertz. A Simple Weight Decay Can Improve Generalization. Advances in Neural Information Processing Systems. 1992 CA:San Mateo,Morgan Kaufmann.
[37] 王文成．神经网络及其在汽车工程中的应用．北京：北京理工大学1998． 7．
[38] 李文兵，田颖．神经元网络技术在连铸漏钢预报中的应用．治金自动化.1998． 3：31-35．
[39] 贾要勤，常炳国．采用神经网络数据融合改善传感器的静态特性．西安交通大学学 报.1999． 11：75-78．
[40] 胡雪萍,刘洪霖．自平衡人工神经网络算法及其在硅钢生产优化中的应用．冶金自动化． 1996． 1：38-40．
[41] Howard Demuth，Mark Beale．Neural Network Toolbox．The MathWorks．Inc．1999． 11．
[42] 汪晓东，万旭，赵鹏程，王骥程．基于神经网络的传感器静态误差综合修正法．仪器仪表学 报．1997，18(3) ：310-313．
[43] J.W.Gardner , P.N.Bartlcll. A brief history of electronic noses. Sensors and Actuators. 1994,19,211-220．
[44] G.Niebling. Identification of gases with classical pattern recognition methods and artificial neural networks. Sensors and Actuators. 1994,19,259-263．
[45] B.S.Barker. Vapour recognition using organic films and artificial neural networks. Sensors and Actuators．1994，17，143-147．
[46] 何芸．运用神经网络进行气体分析的研究．传感器技术 1995，3，12-13．
[47] 庞全，谭炜，杨翠容．智能嗅敏及其实现研究．仪表技术与传感器．1997，7，3-6．
[48] 庞全，谭炜，杨翠容．采用MOS气敏阵列与B-P网络的气体分析方法研究．传感器技 术1997，16(1) ，12-14．
[2] 潘笑，汪小澄．变压器绝缘油击穿电压自动测试仪．变压器.1998，35(4) ：32-34．
[3] 黎洪昌．35-110KV油浸式电流互感器介损测量时应注意的问题．变压器.1999，36(10) : 23-26．
[4] 董旭柱，王昌长，朱德恒．电力变压器局部放电在线监测研究的现状和趋势．变压器．1996，33 (5) ：2-7．
[5] 操敦奎．变压器油中气体分析与诊断，武汉：湖北新生报出版社．1987．
[6] SD187-86． 变压器油中溶解气体分析及判断导则．1986．
[7] 王宗．一种胶囊式储油柜的假油位分析．变压器．1996，33(3) ：39．
[8] 邬家义．国产变压器油发展概述．变压器1996，33(1) ：10-14．
[9] 陆万烈，夏业勤．变压器绕组温度测量的“热模拟”误差．变压器1999，36(10) ：15-17．
[10] 袁道君．油泵故障引起的变压器油金属污染及现场再处理．变压器1999，36(1) ：35-38．
[11] 严璋.电气绝缘在线检测技术．北京：水利电力出版社．1995．
[12] Tsukioka H．New Apparatus for Detecting Transformer Faults．IEEE Trans EI．1986，21(2) ，221．
[13] 李义仓．变压器油中CO和C02与绝缘利料劣化的关系及故障诊断．华东电力1993，21(2) ： 16-19．
[14] 张冠军，钱政，严璋．DGA技术在电力变压器绝缘故障诊断中的应用与进展．变压器1999，36 (1) ：30-34．
[15] 石伟峰．变压器油中气体的TD图分析法．华东电力. 1989，17(1) ：14-18．
[16] 高宁，高文胜，严璋．人工神经网络在电力设备绝缘诊断中应用的进展．高压电器.1997，33 (6) ：31-35．
[17] Lin C E, Ling J M, Huang C L. An Expert System in Transformer Fault Diagnosis Using Dissolved Gas Analysis. IEEE Trans. On Power Deliv. , 1993, 8 (1) .
[18] Tomsovic K, Tapper M, Ingvarsson T. A Fuzzy Information Approach to Integrating Different Transformer Diagnosis Methods. IEEE Trans. On Power Deliv., 1993, 8 (3) ,1638-1646．
[19] Chang C S. Thia B S. Online Rescheduling of Mass Rapid Transit System: Fuzzy Expert System Approach.IEE Proc. Electr. Power Appl. , 1996, 143 (4) , 307-316．
[20] 蔡超豪．利用人工神经网络对变压器故障进行早期诊断．变压器1997，34(1) ：34-37．
[21] 高宁，高文胜.严璋．基于模糊理论与自适应共振网络的油中气体分析诊断．高电压技术 1997，23(4) ：22-25．
[22] Penman J, Yin C M. Feasibility of Using Unsupcrviscd Learning, Artificial Neural Networks for the Condition Monitoring of Electrical Machines. 1EE Proc. Electr. Power Appl , 1994, 141 (6) ,317-322．
[23] McCulloch W S, Pitts W. A Logic Calculus of the Ideal Imminent in Nervous Activity. Bulletin of ath. Biophys. 1943,5:115-133．
[24] Hebb D D. The Organization of Behavior. New York: John Wiley & Sons, 1949．
[25] Rosenblatt F. Principles of Neurodynamics. New York: Spartan Book,1959．
[26] Widrow B, Hoff M E. Adaptive switching circuits. 1960 IRE WESCON Conv. Record. Part 4． 1960,8,96-104．
[27] 王永骥，涂健．神经元网络控制．北京：机械工业出版社．1998，2．
[28] 李孝安，张晓缋．神经网络与神经计算机导论．西安：西北工业大学出版社．1994，10．
[29] Zaromb S, Stetter J. Theoretical basis for identification and measurement of air contaminants using an array of sensors having partly over lapping selectivities. Sensors and Actuators. 1984,6,225-243．
[30] Garey W, et al. Chemometric analysis of multisensor arrays. Sensors and Actuators. 1986,9,223-234．
[31] Muller R, Lange E. Multidimensional sensor for gas analysis. Sensors and Actuators. 1986,9,39-48．
[32] Gall M, Muller R. Investigation of gas mixtures with different MOS gsa sensors with regard to pattern recognition. Sensor and Actuators. 1989,17,583-586．
[33] Hierold C, Muller R. Quantitative analysis of gas mixtures with non-selective gas sensors. Sensors and Actuators. 1989,17,587-592．
[34] Hornor G, Hierold C. Gas analysis by partial modclbuilding. Sensors and Actuators. 1990, B2, 173-184．
[35] R.Hecht-Nielsen. Theory of the Back Propagation Neural Network. Proc of IJCNN, 1989． 1．
[36] Anders Krogh, John A.Hertz. A Simple Weight Decay Can Improve Generalization. Advances in Neural Information Processing Systems. 1992 CA:San Mateo,Morgan Kaufmann.
[37] 王文成．神经网络及其在汽车工程中的应用．北京：北京理工大学1998． 7．
[38] 李文兵，田颖．神经元网络技术在连铸漏钢预报中的应用．治金自动化.1998． 3：31-35．
[39] 贾要勤，常炳国．采用神经网络数据融合改善传感器的静态特性．西安交通大学学 报.1999． 11：75-78．
[40] 胡雪萍,刘洪霖．自平衡人工神经网络算法及其在硅钢生产优化中的应用．冶金自动化． 1996． 1：38-40．
[41] Howard Demuth，Mark Beale．Neural Network Toolbox．The MathWorks．Inc．1999． 11．
[42] 汪晓东，万旭，赵鹏程，王骥程．基于神经网络的传感器静态误差综合修正法．仪器仪表学 报．1997，18(3) ：310-313．
[43] J.W.Gardner , P.N.Bartlcll. A brief history of electronic noses. Sensors and Actuators. 1994,19,211-220．
[44] G.Niebling. Identification of gases with classical pattern recognition methods and artificial neural networks. Sensors and Actuators. 1994,19,259-263．
[45] B.S.Barker. Vapour recognition using organic films and artificial neural networks. Sensors and Actuators．1994，17，143-147．
[46] 何芸．运用神经网络进行气体分析的研究．传感器技术 1995，3，12-13．
[47] 庞全，谭炜，杨翠容．智能嗅敏及其实现研究．仪表技术与传感器．1997，7，3-6．
[48] 庞全，谭炜，杨翠容．采用MOS气敏阵列与B-P网络的气体分析方法研究．传感器技 术1997，16(1) ，12-14．