基于卡尔曼滤波的BP神经网络模型在桥梁形变中的应用
|
摘要
随着我国经济持续稳定的高速发展,高速铁路的修建越来越多。作为关乎国民经济发展的重要基础设施,中国高度重视铁路建设,投资大幅增加。高速铁路要为列车的高速行驶提供一个高平顺性和高稳定性的轨下基础,而桩基作为轨道结构的基础,必须在运营条件下将线路轨道的设计参数保持在要求的标准范围之内,这无疑就对高速铁路的沉降稳定提出了很高的要求。因此,桩基的沉降稳定性以及沉降预测成了高速铁路路基设计和施工的关键。
本文以京沪高铁——昆山段为背景,对桥梁进行定期沉降观测,并对变形数据进行分析处理,并结合当地的地质特征及地质条件,从京沪高铁沉降监测网的建立、观测内容、观测精度、观测频度等方面做比较系统的论述,特别对桩基、桥涵、隧道以及过渡段的沉降观测作了深入研究。
研究了卡尔曼滤波模型和动态神经网络模型在桥梁形变预测中的应用,并在二者的基础上采用了一种基于卡尔曼滤波算法的BP神经网络模型,这种模型能够综合二者的优点,预测精度有了很大的提高。本文中将这种模型应用于桥梁桩基的沉降预测,并取得了较好的效果,为桥梁桩基的沉降预测提供了新的思路。
Continues the stable high speed development along with our country economy, the high-speed railroad construction are more and more many.As concerns the national economy development the important infrastructure, China takes the railroad construction highly, the investment large increase. The high-speed railroad must provide a hing smooth compliance and under the high stable axle for the train high speed travel the foundation, but the pile foundation took the orbital structure the foundation, must maintain under the operation condition the line track design variable in the request standard scope, this without doubt on stably set the very high request to the high-speed railroad subsidence.Therefore, the pile foundation subsidence stability as well as the subsidence forecast the high-speed railroad grade location and the construction key.
This article by Beijing to Shanghai high-valence iron--take Beijing to Shanghai high-valence iron as an example, Periodically to pile settlement observation, and unifies local the geological feature and the geological condition, from Beijing to Shanghai high-valence iron subsidence aspects and so on monitoring network establishment, observation content, observation precision, observation frequency makes the quite systematic elaboration, specially to the pile foundation, the arch of bridge, the tunnel as well as the change-over portion settlement observation has done the thorough research.
Has studied the Kalman filtering model and the dynamic neural network model in bridge deformation forecast application, and used in the two foundation one kind based on the Kalman filtering algorithm BP neural network model, this kind of model can synthesize the two the merit, forecast the precision had the very big enhancement. This will be the model is applied to predict the settlement of pile foundation, bridge and achieved good effect, the settlement of pile foundation for the bridge for new ideas.
本文以京沪高铁——昆山段为背景,对桥梁进行定期沉降观测,并对变形数据进行分析处理,并结合当地的地质特征及地质条件,从京沪高铁沉降监测网的建立、观测内容、观测精度、观测频度等方面做比较系统的论述,特别对桩基、桥涵、隧道以及过渡段的沉降观测作了深入研究。
研究了卡尔曼滤波模型和动态神经网络模型在桥梁形变预测中的应用,并在二者的基础上采用了一种基于卡尔曼滤波算法的BP神经网络模型,这种模型能够综合二者的优点,预测精度有了很大的提高。本文中将这种模型应用于桥梁桩基的沉降预测,并取得了较好的效果,为桥梁桩基的沉降预测提供了新的思路。
Continues the stable high speed development along with our country economy, the high-speed railroad construction are more and more many.As concerns the national economy development the important infrastructure, China takes the railroad construction highly, the investment large increase. The high-speed railroad must provide a hing smooth compliance and under the high stable axle for the train high speed travel the foundation, but the pile foundation took the orbital structure the foundation, must maintain under the operation condition the line track design variable in the request standard scope, this without doubt on stably set the very high request to the high-speed railroad subsidence.Therefore, the pile foundation subsidence stability as well as the subsidence forecast the high-speed railroad grade location and the construction key.
This article by Beijing to Shanghai high-valence iron--take Beijing to Shanghai high-valence iron as an example, Periodically to pile settlement observation, and unifies local the geological feature and the geological condition, from Beijing to Shanghai high-valence iron subsidence aspects and so on monitoring network establishment, observation content, observation precision, observation frequency makes the quite systematic elaboration, specially to the pile foundation, the arch of bridge, the tunnel as well as the change-over portion settlement observation has done the thorough research.
Has studied the Kalman filtering model and the dynamic neural network model in bridge deformation forecast application, and used in the two foundation one kind based on the Kalman filtering algorithm BP neural network model, this kind of model can synthesize the two the merit, forecast the precision had the very big enhancement. This will be the model is applied to predict the settlement of pile foundation, bridge and achieved good effect, the settlement of pile foundation for the bridge for new ideas.
引文
[1]陈国兴,樊良本等.基础工程学.北京：中国水利水电出版社,2002：107-120
[2]高大钊,孙钧.岩土工程的回顾与前瞻．北京：人民交通出版社,2000：160-167
[3]金问鲁，顾尧章.地基基础实用设计施工手册.北京：中国建筑工业出版社,1995：209-210
[4]罗骇先.桩基工程检测手册(第二版).北京：人民交通出版社,2004：1-5
[5]刘金砺,李大展,张雁.桩基设计施工与检测.北京：中国建材工业出版社,2001：13-17
[6]赵明华.土力学与基础工程.武汉：武汉工业大学出版社,2000： 187-193
[7]沈杰.地基基础设计手册.上海：上海科学技术出版社,1988：125-127
[8]姜鹏,张德军.灰色理论模型在路基沉降预测中的应用及其改进[J].山西建筑,2007,32(4)：88-89
[9]Scalero R. S., Tepedelenlioglu N. A Fast Algorithm Train—ing Feedforward Neural Networks. IEEE Trans. On Signal Processing,1992,40(1):202-210
[10].《客运专线铁路无砟轨道铺设条件评估技术指南》(铁建设[2006]158号)；
[11].《客运专线铁路无砟轨道测量技术暂行规定》(铁建设[2006]189号)；
[12].《国家一、二等水准测量规范》(GB12897-2006)；
[13].《建筑沉降变形测量规程》(JGJ／T8-2007)；
[14].《铁路客运专线竣工验收暂行办法》(铁建设[2007]183号)；
[15].《客运专线无砟轨道铁路施工技术指南》(TZ216-2007)；
[16].《工程测量规范》(GB0026-93)；
[17].《全球定位系统(GPS)铁路测量规程》(TB10054-97)；
[18].《客运专线无砟轨道铁路设计指南》(铁建设函[2005]754号)
[19].焦里成．神经网络系统理论.西安：西安交通大学出版社,1990.
[20].夏才初,李永盛.地下工程测试理论与监测技术.上海：同济大学出版社,1999.
[21].史佩栋,高大钊,桂业馄.高层建筑基础工程手册.北京：中国建筑工业出版社,2000.
[22]王科俊；王克成．神经网络建模、预报与控制M,哈尔滨工程大学出版社.1996
[23]陆付民, 顾及时间和水位因子的Kalman滤波法在大坝变形分析中的应用[J], 三峡大学学报,2004,26(5)：3922-3924
[24]刘大杰,于正林,动态测量系统与卡尔曼滤波[J],测绘学报,1988,17(4)：254-262
[25]许国辉,刘跃,卡尔曼滤波的精度评定[J],广州大学学报,2002,2(5)：25-27
[26]梅连友,卡尔曼滤波在滑坡监测中的应用[J],测绘工程,2004,13(3)：13-15
[27]王科俊,金鸿章,战兴群.不同S形函数对BP算法影响的理论及实验研究[C].自动化理论、技术与应用—中国自动化学会第十届青年学术年会论文集西安：西北工业学出版社.299～303,1994
[28]A.G.Parlos,B.Fernandez,A.F.Atiyz,WKTsai. An acceleerated Learning Algorithm for Multi-layer Perception Networks[J].EEE Trans. On NN,1994,5(4):493-497.
[29]Grewal M S and Andrews AP. Kalman Filtering:Theory and Practice[M].New Jersey:Prentice Hall,1993:1～10
[30]Parkum J E,Poulsen N K,and Holst J. Recursive forgetting algorithms[j].International Journal of Control,1992,55(1):109～128.
[31]R. E. Kalman. A New Approach to Linear Filtering and Prediction Theory.ASME. Journal of Basic Eng.1960,82(2):35～46.
[32]R. P. Wisher, J. A. Tabaezynski. A Comparison of Three Nonlinear Filters.Automatica.1969, (5): 457～496.
[33]C. T. Leondes, J. B.Peller, E. B. Stear. Nonlinear Smoothing Theory. IEEETrans. Systems Science and Cybernetics.1970,6(1):63～71.
[34]R. S. Bucy, K. D. Renne. Digital Synthesis of Nonlinear Filters. Automatica.1971,7(3):287～289.
[35]付梦印,邓志红,张继伟Kalman滤波理论及其在导航系统中的应用.北京：科学出版补2003：1-87.
[36]马攀,文鸿雁,离散卡尔曼滤波用于GPS动态变形数据处理[J],桂林工学院学报,2002,22(3)：234
[37]崔希璋,赞宗铸,陶本藻等,广义测量平差(新版)[M],北京：测绘出版社,2001年2月：109-141
[38]杨笃庆,宋迎春,用GPS进行山地滑坡监测的卡尔曼滤波模型[J],湖南人文科技学院学报,2006,3： 5-7
[39]王年生,一种滑坡位移动力学预报方法探讨[J],西部探矿工程,2006,6：269-270；
[2]高大钊,孙钧.岩土工程的回顾与前瞻．北京：人民交通出版社,2000：160-167
[3]金问鲁，顾尧章.地基基础实用设计施工手册.北京：中国建筑工业出版社,1995：209-210
[4]罗骇先.桩基工程检测手册(第二版).北京：人民交通出版社,2004：1-5
[5]刘金砺,李大展,张雁.桩基设计施工与检测.北京：中国建材工业出版社,2001：13-17
[6]赵明华.土力学与基础工程.武汉：武汉工业大学出版社,2000： 187-193
[7]沈杰.地基基础设计手册.上海：上海科学技术出版社,1988：125-127
[8]姜鹏,张德军.灰色理论模型在路基沉降预测中的应用及其改进[J].山西建筑,2007,32(4)：88-89
[9]Scalero R. S., Tepedelenlioglu N. A Fast Algorithm Train—ing Feedforward Neural Networks. IEEE Trans. On Signal Processing,1992,40(1):202-210
[10].《客运专线铁路无砟轨道铺设条件评估技术指南》(铁建设[2006]158号)；
[11].《客运专线铁路无砟轨道测量技术暂行规定》(铁建设[2006]189号)；
[12].《国家一、二等水准测量规范》(GB12897-2006)；
[13].《建筑沉降变形测量规程》(JGJ／T8-2007)；
[14].《铁路客运专线竣工验收暂行办法》(铁建设[2007]183号)；
[15].《客运专线无砟轨道铁路施工技术指南》(TZ216-2007)；
[16].《工程测量规范》(GB0026-93)；
[17].《全球定位系统(GPS)铁路测量规程》(TB10054-97)；
[18].《客运专线无砟轨道铁路设计指南》(铁建设函[2005]754号)
[19].焦里成．神经网络系统理论.西安：西安交通大学出版社,1990.
[20].夏才初,李永盛.地下工程测试理论与监测技术.上海：同济大学出版社,1999.
[21].史佩栋,高大钊,桂业馄.高层建筑基础工程手册.北京：中国建筑工业出版社,2000.
[22]王科俊；王克成．神经网络建模、预报与控制M,哈尔滨工程大学出版社.1996
[23]陆付民, 顾及时间和水位因子的Kalman滤波法在大坝变形分析中的应用[J], 三峡大学学报,2004,26(5)：3922-3924
[24]刘大杰,于正林,动态测量系统与卡尔曼滤波[J],测绘学报,1988,17(4)：254-262
[25]许国辉,刘跃,卡尔曼滤波的精度评定[J],广州大学学报,2002,2(5)：25-27
[26]梅连友,卡尔曼滤波在滑坡监测中的应用[J],测绘工程,2004,13(3)：13-15
[27]王科俊,金鸿章,战兴群.不同S形函数对BP算法影响的理论及实验研究[C].自动化理论、技术与应用—中国自动化学会第十届青年学术年会论文集西安：西北工业学出版社.299～303,1994
[28]A.G.Parlos,B.Fernandez,A.F.Atiyz,WKTsai. An acceleerated Learning Algorithm for Multi-layer Perception Networks[J].EEE Trans. On NN,1994,5(4):493-497.
[29]Grewal M S and Andrews AP. Kalman Filtering:Theory and Practice[M].New Jersey:Prentice Hall,1993:1～10
[30]Parkum J E,Poulsen N K,and Holst J. Recursive forgetting algorithms[j].International Journal of Control,1992,55(1):109～128.
[31]R. E. Kalman. A New Approach to Linear Filtering and Prediction Theory.ASME. Journal of Basic Eng.1960,82(2):35～46.
[32]R. P. Wisher, J. A. Tabaezynski. A Comparison of Three Nonlinear Filters.Automatica.1969, (5): 457～496.
[33]C. T. Leondes, J. B.Peller, E. B. Stear. Nonlinear Smoothing Theory. IEEETrans. Systems Science and Cybernetics.1970,6(1):63～71.
[34]R. S. Bucy, K. D. Renne. Digital Synthesis of Nonlinear Filters. Automatica.1971,7(3):287～289.
[35]付梦印,邓志红,张继伟Kalman滤波理论及其在导航系统中的应用.北京：科学出版补2003：1-87.
[36]马攀,文鸿雁,离散卡尔曼滤波用于GPS动态变形数据处理[J],桂林工学院学报,2002,22(3)：234
[37]崔希璋,赞宗铸,陶本藻等,广义测量平差(新版)[M],北京：测绘出版社,2001年2月：109-141
[38]杨笃庆,宋迎春,用GPS进行山地滑坡监测的卡尔曼滤波模型[J],湖南人文科技学院学报,2006,3： 5-7
[39]王年生,一种滑坡位移动力学预报方法探讨[J],西部探矿工程,2006,6：269-270；