基于人工神经网络的城市燃气管道的土壤腐蚀性评价研究
|
摘要
随着我国现代化进程的不断加快,人民生活水平的提高,以及全球环境问题的日益突出,节能减排压力越来越大,天然气的需求呈爆炸式增长,快速地推动了城市燃气输气管网的建设。城市燃气输配系统通常由钢制管道埋地建设而成,因而,钢制埋地钢制管道的腐蚀与防护便成了急待解决的问题。作为钢制管道腐蚀的重要因素之一——土壤腐蚀也一直被国内外研究者所关注。但是至今仍没有一个大家公认的,比较切合实际的土壤腐蚀性评价方法。
本文在总结和比较了前人针对埋地钢制管道土壤腐蚀性评价研究方法的基础上,采用神经网络法建立了土壤腐蚀性评价模型,该模型具有2个重要特点:一是简化了样本检测参数。从训练样本的来源上,不需要测试大量的土壤理化性质参数,而是需要通过综合影响管道土壤腐蚀的各主要因素土壤电阻率、氧化还原电位、含盐量、含水量、PH值)的单项指标,就可以建立新的多项综合指标来建立模型;其二是增加了训练样本。从训练样本的数量上,为了提高模型的鲁棒性和识别的准确性,对样本进行了扩充,在每级标准之间随机生成40组训练样本。
本文在新的理论基础上,将建立的BP和RBF两种网络运用到廊坊市燃气管道的腐蚀调查中,应用结果表明,RBF网络在对新样本的预测精度方面优于BP网络。该方法可用于现场管道的土壤腐蚀评价,同时也可以为管道的及时维修和更换提供科学的依据。
With the accelerating pace of modernization of our country, the improvement of people’s living standards and the increasingly outstanding global environmental problems, the pressure of energy-saving and environmental protection is more and more big. The explosive growth of natural gas demand is promoting the construction of urban gas network rapidly. Urban gas transmission and distribution system is usually constructed by buried steel pipe, so the corrosion and protection of steel pipeline has become problems to be solved. As one of the important factors—soil erosion is always concerned by researchers at home and abroad. But up to now, we havn’t find a recognized and more realistic evaluation of soil corrosion.
Based on the evaluation method of pre-scholars, this paper established the model of soil corrosion evaluation by ANN. This model has two important features:the first, it simplifies the testing samples parameters. It doesn’t need a lot of soil physical and chemical properties, but only need combine some important individual index of soil pipe corrosion(soil resistivity, oxidation—reduction potential, salinity, water content, PH), we can create a new integrated model. The second, it increases the number of training samples. In order to improve the robustness of the model and the accuracy of the identification, the samples were expanded by generating 40 data sets randomly in each class standards.
On the basis of new theory, this paper used BP and RBF network into corrosion survey of gas pipelines in Langfang. The application results show that the RBF network is better than BP network in the forecast accuracy to new samples. This method not only can be used for on-site assessment of soil corrosion of urban gas pipeline, but also provide a scientific basis for urban gas pipelines’maintenance and timely replacement.
本文在总结和比较了前人针对埋地钢制管道土壤腐蚀性评价研究方法的基础上,采用神经网络法建立了土壤腐蚀性评价模型,该模型具有2个重要特点:一是简化了样本检测参数。从训练样本的来源上,不需要测试大量的土壤理化性质参数,而是需要通过综合影响管道土壤腐蚀的各主要因素土壤电阻率、氧化还原电位、含盐量、含水量、PH值)的单项指标,就可以建立新的多项综合指标来建立模型;其二是增加了训练样本。从训练样本的数量上,为了提高模型的鲁棒性和识别的准确性,对样本进行了扩充,在每级标准之间随机生成40组训练样本。
本文在新的理论基础上,将建立的BP和RBF两种网络运用到廊坊市燃气管道的腐蚀调查中,应用结果表明,RBF网络在对新样本的预测精度方面优于BP网络。该方法可用于现场管道的土壤腐蚀评价,同时也可以为管道的及时维修和更换提供科学的依据。
With the accelerating pace of modernization of our country, the improvement of people’s living standards and the increasingly outstanding global environmental problems, the pressure of energy-saving and environmental protection is more and more big. The explosive growth of natural gas demand is promoting the construction of urban gas network rapidly. Urban gas transmission and distribution system is usually constructed by buried steel pipe, so the corrosion and protection of steel pipeline has become problems to be solved. As one of the important factors—soil erosion is always concerned by researchers at home and abroad. But up to now, we havn’t find a recognized and more realistic evaluation of soil corrosion.
Based on the evaluation method of pre-scholars, this paper established the model of soil corrosion evaluation by ANN. This model has two important features:the first, it simplifies the testing samples parameters. It doesn’t need a lot of soil physical and chemical properties, but only need combine some important individual index of soil pipe corrosion(soil resistivity, oxidation—reduction potential, salinity, water content, PH), we can create a new integrated model. The second, it increases the number of training samples. In order to improve the robustness of the model and the accuracy of the identification, the samples were expanded by generating 40 data sets randomly in each class standards.
On the basis of new theory, this paper used BP and RBF network into corrosion survey of gas pipelines in Langfang. The application results show that the RBF network is better than BP network in the forecast accuracy to new samples. This method not only can be used for on-site assessment of soil corrosion of urban gas pipeline, but also provide a scientific basis for urban gas pipelines’maintenance and timely replacement.
引文
[1]张静.天然气与中国能源发展前景[J].中国石油和化工经济分析,2007,(12): 20-24.
[2]孙岩冰.纵贯四方,“气”势待发[J].中国石油石化,2008,(3):17-20.
[3]吴汮.国内外土壤腐蚀研究工作概况[J].材料保护,1983,5(32):32-35.
[4]托马晓夫著,华宝定译.金属腐蚀及其保护理论[M].北京:中国工业出版社,1964.
[5] Romanoff.M., Underground corrosion[M], NBS Circalar,1957.
[6] Gupta,B.K,etal. Science and Technology[M],1976.
[7]郭晋.华南酸性土壤对碳钢腐蚀性的研究[D].广州:华南理工大学工学, 2000.
[8]中科院应用化学所.应用化学集刊[M],1961,(5):55.
[9] SY/T0087-1995.钢质管道及储罐腐蚀与防护调查方法标准[S].北京:石油工业出社,2005.
[10]美国腐蚀工程师协会编,朱日彰等译.腐蚀与防护技术基础[M].北京:冶金工业出社,1987.
[11]国家科学技术委员会,国家自然科学基金委员会,全国土壤腐蚀试验网站编.材料土壤腐蚀试验方法[M].北京:科学出版社,1990.
[12] Romanoff.M., Underground Corrosion. National Technical Information Service(NTIS)[M], U.S Department of Commerce. April,l957.
[13]托马晓夫著,华宝定译.金属腐蚀及其保护理论[M].北京:中国工业出版社,1964.
[14] U.R.艾万思著,华保定译.金属的腐蚀与氧化[M].北京,机械工业出版社,1976.
[15] Waters F.0.Corrosion[M]. 1952.
[16] Booth G.H. CooperA.W and Cooper P.M .Corrosion[M].1967.
[17] Abbot W.K ,Schillmollr C.M. Metal Protect[M].1963.
[18] Kasahara K,Adachi H and Kajiyama F.8th International Congression on Metallic Corrosion(9th ICMC)[C].1981:Vol.2:1832.
[19]叶康民著.金属腐蚀与防护概论[M].北京:人民教育出版社,1981.
[20] Gimelfarb M .N.Materials Performance[J].1990,29(2)18-20.
[21] AshworthV,GooganC.G and JacobW.R, Australasian Corrosion Association Inc,Proceedings Conference 26[C].South Australia:l986: Vol.2:1.
[22]金属腐蚀组编.中国科学院长春应用化学研究所集刊[C].第5集,1961:55.
[23]王强.地下金属管道的腐蚀与阴极保护[M],青海人民出版社,1984:39.
[24]华东输油管理局.土壤腐蚀数据处理及腐蚀性综合评价鉴定资料[C].徐州.1991.
[25]中国腐蚀与防护学会金属腐蚀手册编辑委员会主编.金属腐蚀手册[M].上海:上海科学技术出版社,1987.
[26] Brerkeley K.G.C. Corrosion control for buried pipelines Guides to Practice[J]. No 10,Dept of Industry.London.1982.
[27] Code of practice for cathodic protection[S]. CP1021:1973. British Standards Institution. London.1973.
[28]李成保,周关明.用极化电阻RP评价土壤对铅腐蚀的研究[J].材料保护.1981,(3):7-9.
[29] Beackman W V著,赖敬文译.阴极保护简明手册[M].北京:石油工业出版社,1987:90.
[30]美国国家标准[S].ANSI/AWWAC105,A21,5.
[31]李双林,杨志勇等.因子分析法在土壤腐蚀性研究中的应用[J].腐蚀科学与防护技术,1995,7(3):263-265.
[32]张扬,李帆等.埋地燃气钢管土壤腐蚀性模糊综合评价[J].煤气与热力,2006,26(4):12-14
[33]翁永基,李相怡等.金属土壤腐蚀模型中关键参数选择的研究[J].腐蚀科学与防护技术,1997,9(4): 281-285.
[34]李余斌,黄坤等.基于神经网络的管道沿线土壤腐蚀态势评价[J].油气运,2007,26(8),47-49.
[35] Simon Haykin著,叶世伟、史忠植译.神经网络原理[M].北京:机械工业出版社,2004.
[36]周开利,康耀红.神经网络模型及其MATLAB仿真程序设计[M].北京:清华大学出版社,2005.
[37]魏海坤编著.神经网络结构设计的理论与方法[M].北京:国防工业出版社,2005.
[38]王伟著.人工神经元网络原理:入门与应用[M].北京:北京航空航天大学出版社,1995.
[39]张立明.人工神经元网络的模型及其运用[M].上海:复旦大学出版社,1993.
[40]飞思科技产品研发中心. Matlab6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003.
[41]耿小庆,和金生,于宝琴.几种改进BP算法及其在应用中的比较分析[M].计算机工程与应用,2007,43(33):243-245.
[42]苏高利,邓芳萍.论基于MATLAB语言的BP神经网络的改进算法[M].科技通报,2003,19(2):130-135.
[43] Moller M.F.A scaled conjugate gradient algorithm for fast supervised learning[J].Neural Networks,vol.6, 525-533,1993.
[44] Dennis J.E,Schnabe R.B.1.Numerrical Methods for Unconstrained Optimization and Nonlinear Equations.Englewood Cliffs[J],NJ:Prentice-Hall,1983.
[45] Battiti R. First and second order methods for learning: Between steepest descent and Newton’s method [J ]. Neural Computation,1992 , 4(2):141-166.
[46] Ackley D.H, Hinton G..F,Sejnowski T.J.A learning algorithm for Boltzman machines. Cognitive Science[J],vol.9,147-169,1985.
[47] Moody J, Darken C. Learning with the Localized Receptive Fields In Proc Connectionist Models Summer School[M], Camegie Mellon Univ. 1988.
[48]周俊武,孙传尧,王福利.径向基函数(RBF)网络的研究及实现[J].矿冶,2001,(10):71-72.
[49] Moody J, Darken C. Fast Learning in Networks of Locally-tuned Processing Units. Neural Computation[J], 1989,(1):281-294.
[50]周勇,胡中功.RBF神经网络理论及其在控制中的应用[J].武汉科技学院学报,2007,20(5) : 40-43.
[51]桑松,林焰,纪卓尚.基于神经网络的船型要素数学建模研究[J].计算机工程,2002,28 (9) 238-239.
[52] Moody J E. The effective number of parameters:an analysis of generalization and regularization in nonlinear learning system [A].In :Advances in Neural Information Processing Systems[C] .San Mateo,1992, 4 :847-854
[53] Chen S,Cowan C F N,Grant P N. Orthogonal Least squares learning algorithms for radial basis function networks [J],IEEE Trans .Neural Networks,1991, 2(2) :302-309
[54]张志涌.掌握和精通MATLAB[M].北京:北京航空航天大学出版社,1997
[55] Demuth, Howard B. Neural network design[M]. Beijing :China Machine Press,2002.
[56]姜绍飞.人工神经网络用于建筑工程领域的数据处理方法[M].哈尔滨建筑大学学报,1999(5):24-32.
[57]董长虹.MATLAB神经网络与应用[M].北京:国防工业出版社,2005.
[58] Lippmann R P. An introduction to computer with neural nets[J]]. IEEE ASSP Magazine, 1987(4):4—22
[59] Necht Nielsen. Counter propagation Networks[M]. Applied Optics.1987, 26(12) : 4979-4984
[60]张立明.人工神经网络的模型及应用[M].上海:复旦大学出版社,1993.
[2]孙岩冰.纵贯四方,“气”势待发[J].中国石油石化,2008,(3):17-20.
[3]吴汮.国内外土壤腐蚀研究工作概况[J].材料保护,1983,5(32):32-35.
[4]托马晓夫著,华宝定译.金属腐蚀及其保护理论[M].北京:中国工业出版社,1964.
[5] Romanoff.M., Underground corrosion[M], NBS Circalar,1957.
[6] Gupta,B.K,etal. Science and Technology[M],1976.
[7]郭晋.华南酸性土壤对碳钢腐蚀性的研究[D].广州:华南理工大学工学, 2000.
[8]中科院应用化学所.应用化学集刊[M],1961,(5):55.
[9] SY/T0087-1995.钢质管道及储罐腐蚀与防护调查方法标准[S].北京:石油工业出社,2005.
[10]美国腐蚀工程师协会编,朱日彰等译.腐蚀与防护技术基础[M].北京:冶金工业出社,1987.
[11]国家科学技术委员会,国家自然科学基金委员会,全国土壤腐蚀试验网站编.材料土壤腐蚀试验方法[M].北京:科学出版社,1990.
[12] Romanoff.M., Underground Corrosion. National Technical Information Service(NTIS)[M], U.S Department of Commerce. April,l957.
[13]托马晓夫著,华宝定译.金属腐蚀及其保护理论[M].北京:中国工业出版社,1964.
[14] U.R.艾万思著,华保定译.金属的腐蚀与氧化[M].北京,机械工业出版社,1976.
[15] Waters F.0.Corrosion[M]. 1952.
[16] Booth G.H. CooperA.W and Cooper P.M .Corrosion[M].1967.
[17] Abbot W.K ,Schillmollr C.M. Metal Protect[M].1963.
[18] Kasahara K,Adachi H and Kajiyama F.8th International Congression on Metallic Corrosion(9th ICMC)[C].1981:Vol.2:1832.
[19]叶康民著.金属腐蚀与防护概论[M].北京:人民教育出版社,1981.
[20] Gimelfarb M .N.Materials Performance[J].1990,29(2)18-20.
[21] AshworthV,GooganC.G and JacobW.R, Australasian Corrosion Association Inc,Proceedings Conference 26[C].South Australia:l986: Vol.2:1.
[22]金属腐蚀组编.中国科学院长春应用化学研究所集刊[C].第5集,1961:55.
[23]王强.地下金属管道的腐蚀与阴极保护[M],青海人民出版社,1984:39.
[24]华东输油管理局.土壤腐蚀数据处理及腐蚀性综合评价鉴定资料[C].徐州.1991.
[25]中国腐蚀与防护学会金属腐蚀手册编辑委员会主编.金属腐蚀手册[M].上海:上海科学技术出版社,1987.
[26] Brerkeley K.G.C. Corrosion control for buried pipelines Guides to Practice[J]. No 10,Dept of Industry.London.1982.
[27] Code of practice for cathodic protection[S]. CP1021:1973. British Standards Institution. London.1973.
[28]李成保,周关明.用极化电阻RP评价土壤对铅腐蚀的研究[J].材料保护.1981,(3):7-9.
[29] Beackman W V著,赖敬文译.阴极保护简明手册[M].北京:石油工业出版社,1987:90.
[30]美国国家标准[S].ANSI/AWWAC105,A21,5.
[31]李双林,杨志勇等.因子分析法在土壤腐蚀性研究中的应用[J].腐蚀科学与防护技术,1995,7(3):263-265.
[32]张扬,李帆等.埋地燃气钢管土壤腐蚀性模糊综合评价[J].煤气与热力,2006,26(4):12-14
[33]翁永基,李相怡等.金属土壤腐蚀模型中关键参数选择的研究[J].腐蚀科学与防护技术,1997,9(4): 281-285.
[34]李余斌,黄坤等.基于神经网络的管道沿线土壤腐蚀态势评价[J].油气运,2007,26(8),47-49.
[35] Simon Haykin著,叶世伟、史忠植译.神经网络原理[M].北京:机械工业出版社,2004.
[36]周开利,康耀红.神经网络模型及其MATLAB仿真程序设计[M].北京:清华大学出版社,2005.
[37]魏海坤编著.神经网络结构设计的理论与方法[M].北京:国防工业出版社,2005.
[38]王伟著.人工神经元网络原理:入门与应用[M].北京:北京航空航天大学出版社,1995.
[39]张立明.人工神经元网络的模型及其运用[M].上海:复旦大学出版社,1993.
[40]飞思科技产品研发中心. Matlab6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003.
[41]耿小庆,和金生,于宝琴.几种改进BP算法及其在应用中的比较分析[M].计算机工程与应用,2007,43(33):243-245.
[42]苏高利,邓芳萍.论基于MATLAB语言的BP神经网络的改进算法[M].科技通报,2003,19(2):130-135.
[43] Moller M.F.A scaled conjugate gradient algorithm for fast supervised learning[J].Neural Networks,vol.6, 525-533,1993.
[44] Dennis J.E,Schnabe R.B.1.Numerrical Methods for Unconstrained Optimization and Nonlinear Equations.Englewood Cliffs[J],NJ:Prentice-Hall,1983.
[45] Battiti R. First and second order methods for learning: Between steepest descent and Newton’s method [J ]. Neural Computation,1992 , 4(2):141-166.
[46] Ackley D.H, Hinton G..F,Sejnowski T.J.A learning algorithm for Boltzman machines. Cognitive Science[J],vol.9,147-169,1985.
[47] Moody J, Darken C. Learning with the Localized Receptive Fields In Proc Connectionist Models Summer School[M], Camegie Mellon Univ. 1988.
[48]周俊武,孙传尧,王福利.径向基函数(RBF)网络的研究及实现[J].矿冶,2001,(10):71-72.
[49] Moody J, Darken C. Fast Learning in Networks of Locally-tuned Processing Units. Neural Computation[J], 1989,(1):281-294.
[50]周勇,胡中功.RBF神经网络理论及其在控制中的应用[J].武汉科技学院学报,2007,20(5) : 40-43.
[51]桑松,林焰,纪卓尚.基于神经网络的船型要素数学建模研究[J].计算机工程,2002,28 (9) 238-239.
[52] Moody J E. The effective number of parameters:an analysis of generalization and regularization in nonlinear learning system [A].In :Advances in Neural Information Processing Systems[C] .San Mateo,1992, 4 :847-854
[53] Chen S,Cowan C F N,Grant P N. Orthogonal Least squares learning algorithms for radial basis function networks [J],IEEE Trans .Neural Networks,1991, 2(2) :302-309
[54]张志涌.掌握和精通MATLAB[M].北京:北京航空航天大学出版社,1997
[55] Demuth, Howard B. Neural network design[M]. Beijing :China Machine Press,2002.
[56]姜绍飞.人工神经网络用于建筑工程领域的数据处理方法[M].哈尔滨建筑大学学报,1999(5):24-32.
[57]董长虹.MATLAB神经网络与应用[M].北京:国防工业出版社,2005.
[58] Lippmann R P. An introduction to computer with neural nets[J]]. IEEE ASSP Magazine, 1987(4):4—22
[59] Necht Nielsen. Counter propagation Networks[M]. Applied Optics.1987, 26(12) : 4979-4984
[60]张立明.人工神经网络的模型及应用[M].上海:复旦大学出版社,1993.