基于BP人工神经网络的基础隔震智能初步设计系统研究
|
摘要
基础隔震技术作为一种有效的抗震设计手段在我国得到广泛的应用,为减轻地震灾害提供了一条合理、有效、安全的新途径,并在防震减灾事业中起到积极的推动作用。2008年汶川地震后,隔震结构在地震中表现出良好的抗震性能,使得隔震技术在我国得到了更加迅速的发展。
针对隔震结构发展迅速且设计方法比较复杂这一现状,本文利用人工神经网络建立以隔震设计样本集为基础的因果关系网络,通过神经网络反映出上部结构基本参数与隔震后结构最大层剪力比和隔震支座最大位移之间较强的非线性关系,提出一种快速、高效的基于BP神经网络的基础隔震智能初步设计系统。
首先,应用标准BP人工神经网络,以建筑结构的抗震设防类别、设防烈度、场地类别、地震分组、高宽比、长宽比、刚度,质量和面积为主要影响因子,以隔震后结构的最大层剪力比和支座最大位移为输出结果,建立隔震初步设计系统。通过网络测试表明该网络的准确率较高,说明基于标准BP神经网络的隔震初步设计系统对隔震结构减震效果的分析具有一定的可行性,但在网络的训练速度、精度等方面还有待进一步提高。
其次,针对基于标准BP神经网络的基础隔震初步设计系统存在的问题,对标准BP神经网络进行改进,改进后的BP神经网络有效的提高了网络的学习速度,缩短了训练时间,提高了网络测试准确率。说明本文采用的改进方法是可行的,改进后的神经网络具有良好得推广价值和实际应用价值。
最后,应用改进后的BP神经网络,结合隔震结结构设计的特点,建立了基础隔震智能设计系统。并利用MATLAB7.5强大的图形处理功能,设计了本系统的操作界面,使整个基础隔震智能初步设计系统具有良好的实用性和交互性。
通过工程实例验证,本文建立的基础隔震智能初步设计系统操作简单、使用方便,能够将前人的设计经验通过神经网络储存并推广,为基础隔震设计起到辅助的作用。从而极大地提高工程设计的质量,缩短设计周期、加快设计进程,实现经验共享,提高整个隔震设计效率。
As an efficient method of seismic design, base-isolation technology has been widely applied in our country, it provides a new approach reasonably which is effective and safe to the reduction of earthquake disasters and plays an active driving role in the works of earthquake administration. After Wen-chuan earthquake in 2008, base-isolation structure had good seismic performance in this earthquake, it more rapidly develop in our country.
Aiming at the present situation of rapidly develop of base-isolation structure and complex design method, using artificial neural network established causality network based on base-isolation design samples, reflects nonlinear relation between basic parameters of superstructure and the maximum shear force ratio and the largest displacement of isolation bearing, the paper proposed a kind of rapid and efficient base-isolation intelligent initial design system based on BP neural network.
First, using standard BP neural network, considering the seismic fortification type, fortification intensity, site type, seismic design group, height-width ratio, aspect ratio, stiffness, mass and area of structure as main impact factor, as the maximum shear force ratio and the largest displacement of isolation bearing to output data, this article sets up a base isolation structural initial design system. It shows that high accuracy by network tested and indicates that the analysis of vibration-isolating effect in base-isolation structure by base-isolation initial design system based on standard BP neural network has certain feasibility, but training speed and accuracy of network should be further improved.
Second, aiming at the existing problems of base-isolation initial design system based on standard BP neural network, this paper discusses the existing defects of standard BP neural network and improves of these defects. Improved neural network improves the learning speed and the prediction accuracy, reduces the training time of the network. These indicate that the improved method is feasible and the improved BP neural network has a good application and promotional value.
Finally, applying the improved BP neural network, and combined with the characteristics of base-isolation design, base-isolation intelligent initial design system can be established and graphical user interface can be make by MATLAB7.5 programming, which will make the system better practicality and interaction.
Through the engineering practice, the base-isolation intelligent initial design system in this paper is simple to operate and easy to use and could store and extend previous experience as an auxiliary function for base-isolation design. Thus it can improve the quality of engineering design greatly, shorten design period, quickening design process, sharing experiences and improve design efficiency.
针对隔震结构发展迅速且设计方法比较复杂这一现状,本文利用人工神经网络建立以隔震设计样本集为基础的因果关系网络,通过神经网络反映出上部结构基本参数与隔震后结构最大层剪力比和隔震支座最大位移之间较强的非线性关系,提出一种快速、高效的基于BP神经网络的基础隔震智能初步设计系统。
首先,应用标准BP人工神经网络,以建筑结构的抗震设防类别、设防烈度、场地类别、地震分组、高宽比、长宽比、刚度,质量和面积为主要影响因子,以隔震后结构的最大层剪力比和支座最大位移为输出结果,建立隔震初步设计系统。通过网络测试表明该网络的准确率较高,说明基于标准BP神经网络的隔震初步设计系统对隔震结构减震效果的分析具有一定的可行性,但在网络的训练速度、精度等方面还有待进一步提高。
其次,针对基于标准BP神经网络的基础隔震初步设计系统存在的问题,对标准BP神经网络进行改进,改进后的BP神经网络有效的提高了网络的学习速度,缩短了训练时间,提高了网络测试准确率。说明本文采用的改进方法是可行的,改进后的神经网络具有良好得推广价值和实际应用价值。
最后,应用改进后的BP神经网络,结合隔震结结构设计的特点,建立了基础隔震智能设计系统。并利用MATLAB7.5强大的图形处理功能,设计了本系统的操作界面,使整个基础隔震智能初步设计系统具有良好的实用性和交互性。
通过工程实例验证,本文建立的基础隔震智能初步设计系统操作简单、使用方便,能够将前人的设计经验通过神经网络储存并推广,为基础隔震设计起到辅助的作用。从而极大地提高工程设计的质量,缩短设计周期、加快设计进程,实现经验共享,提高整个隔震设计效率。
As an efficient method of seismic design, base-isolation technology has been widely applied in our country, it provides a new approach reasonably which is effective and safe to the reduction of earthquake disasters and plays an active driving role in the works of earthquake administration. After Wen-chuan earthquake in 2008, base-isolation structure had good seismic performance in this earthquake, it more rapidly develop in our country.
Aiming at the present situation of rapidly develop of base-isolation structure and complex design method, using artificial neural network established causality network based on base-isolation design samples, reflects nonlinear relation between basic parameters of superstructure and the maximum shear force ratio and the largest displacement of isolation bearing, the paper proposed a kind of rapid and efficient base-isolation intelligent initial design system based on BP neural network.
First, using standard BP neural network, considering the seismic fortification type, fortification intensity, site type, seismic design group, height-width ratio, aspect ratio, stiffness, mass and area of structure as main impact factor, as the maximum shear force ratio and the largest displacement of isolation bearing to output data, this article sets up a base isolation structural initial design system. It shows that high accuracy by network tested and indicates that the analysis of vibration-isolating effect in base-isolation structure by base-isolation initial design system based on standard BP neural network has certain feasibility, but training speed and accuracy of network should be further improved.
Second, aiming at the existing problems of base-isolation initial design system based on standard BP neural network, this paper discusses the existing defects of standard BP neural network and improves of these defects. Improved neural network improves the learning speed and the prediction accuracy, reduces the training time of the network. These indicate that the improved method is feasible and the improved BP neural network has a good application and promotional value.
Finally, applying the improved BP neural network, and combined with the characteristics of base-isolation design, base-isolation intelligent initial design system can be established and graphical user interface can be make by MATLAB7.5 programming, which will make the system better practicality and interaction.
Through the engineering practice, the base-isolation intelligent initial design system in this paper is simple to operate and easy to use and could store and extend previous experience as an auxiliary function for base-isolation design. Thus it can improve the quality of engineering design greatly, shorten design period, quickening design process, sharing experiences and improve design efficiency.
引文
[1]周福霖.工程结构减震控制[M].北京：地震出版社,1997,4-56
[2]侯鹏.夹层橡胶基础隔震结构动力分析[J].成都：西南交通大学硕士学位论文,2003,1-2
[3]日本免震构造协会编,叶列平译.隔震结构入门.北京：科学出版社,1998,4-10
[4]李慧,杜永峰,屠锦敏,周茗如,邓学晶.叠层橡胶基础隔震技术的研究[J].甘肃工业大学学报,1999,25(2)：81-83
[5]李慧,杜永峰,屠锦敏,梁文钊,黎海南,王光照.兰州市某住宅楼工程抗震方法的优化选择[J].基建优化,2000,21(2)：30-32
[6]王社良.抗震结构设计[M].武汉：武汉理工大学出版社,2008,224-232
[7]Farzad Naeim, James M.Kelly. Design of seismic isolated structures[M]. canda, John, wiley & Sons Inc,1999,35-46
[8]GB 50011-2001,建筑抗震设计规范[S]
[9]CECS126:2001叠层橡胶支座隔震技术规程[S]
[10]日本建筑学会编,刘文光译.隔震结构设计[M].北京：地震出版社,2006,28-40
[11]张志公.大底盘多塔楼结构基础隔震性能分析与研究[M].兰州理工大学硕士学位
论文,23-30
[12]CECS160:2004建筑工程抗震性态设计通则(试用)[S]
[13]党育,杜永峰,李慧.基础隔震结构设计及施工指南[M].北京：中国水利水电出版社,2007,73-79
[14]罗四维.神经网络建造[M].北京：中国铁道出版社,1998,3-5
[15]赵林明,胡浩云,魏德华,王树谦.多层前向人工神经网络.郑州：黄河水利出版社,1999,69-90
[16]Ghaboussi.J,Joghataie.A.Active control of structures using neural networks [J]. Eng. Mec,1995,4(55):554-567
[17]魏东.非线性系统神经网络参数预测及控制[M].北京：机械工业出版社,2008,34-38
[18]张代远.神经网络新理论与方法[M].北京：清华大学出版社,2006,4-5
[19]韩立群.人工神经网络教程[M].北京：北京邮电大学出版社,2006,63-75
[20]刘伯权,王经建.基于神经网络的地震区结构经济性分析系统[J].长安大学学报,2004,24(5)：55-59
[21]黄浩,陶龙光,李幻.人工神经网络在岩土工程中的应用状况及前景[J].工业建筑,2007,37(增刊)：844-847
[22]汤皓,陈国兴.灰关联与人工神经网络在建筑物震害预测中的应用[J].地震工程与工 程振动,2006,26(3)：57-59
[23]梅青海,霍凯成.基于人工神经网络的桩基础选型研究[J].武汉理工大学学报,2004,26(12)：63-65
[24]郑浩,王全风.BP神经网络在高层结构选型中的应用[J].华侨大学学报,2003,24(1)：48-55
[25]王新刚,王起才,周岩.基于人工神经网络的大体积混凝土温度场预测[J].混凝土2006,203(9)：21-24
[26]李慧,王红丽,杜永峰.钢框架连接损伤的两种改进BP算法的比较[J].工程抗震与加固改造,2006,28(6)：39-42
[27]李慧,王红丽,杜永峰,张东兵.基于改进BP算法的刚框架连接损伤识别方法[J].甘肃科学学报,2007,19(2)：114-117
[28]杜永峰,李慧.高效神经网络训练及其在桁架损伤识别中的应用[J].华中科技大学学报,2008,25(4)：51-53
[29]杜永峰,郭剑虹.基于RBF神经网络的结构动力响应预测[J].兰州理工大学学报,2006,32(2)：111-114
[30]赵艳林,杨军平,黄剑飞,吕海波.基于BP神经网络的单层钢筋混凝土柱工业厂房震害预测[J].桂林工学院,2006,26(4)：491-496
[31]Qin Zheng,Wuzhong fu,Xiaofeng Liao,et al. A network Intrusion Detection Architecture Based on Intelligent Agents.Chong qin:computer science.2005,32(6):133-136
[32]韩立群.人工神经网络理论、设计及应用[M].北京：化学工业出版社,2002,44-56
[33]许东,吴铮.基于MATLAB 6.X的系统分析与设计-神经网络[M].西安：西安电子科技大学出版社,2002,19-24
[34]飞思科技产品研发中心.MATLAB 6.5辅助神经网络分析与设计[M].北京：电子工业出版社,2003,64-69
[35]Lee S.C,Heinbuch D.V.Training a neural-network based intrusion detector to recongnize novel attacks[J].IEEE Transactions on Systems.2001,31(4):294-299
[36]魏海坤.神经网络结构设计的理论与方法[M].北京：国防工业出版社,2007,36-39
[37]Sung A H,Mukkamala S.Identifying Important Features for Intrusion Detection Using Support Vector Machines and Neural Networks. Applications and the Internet Proceedings[J].USA:2003 symposium on Publication,2003,209-216
[38]Boozarjomehry,R.B.Svrcek,W.Y. Automatic design of neural network structures [J]. Computers and Chemical Engineering,2000,25 (7-8):1075-1088
[39]王红丽.框架结构节点损伤识别的数值与实验研究[M].兰州理工大学硕士学位论文,2007,13-27
[40]Ignizio, James P, Soltys, James R. Simultaneous design and training of ontogenic neural network classifiers[J].Computers & Operations Research,1999,23 (6):535-546
[41]何雪忠.复杂大分子体系平衡性质的神经网络预测[M].郑州：郑州大学硕士学位论文,2004,37-39
[42]王燚,郭伟.一种混沌神经网络的混沌时间序列预测[J].电子测量与仪器学报,2007,21(6)：5-9
[43]Sarkav D.Methods to speed up error BP learning aloritmn[J]. ACM Computing survey, 1995,27(3):519-592
[44]杜刚.改进的BP神经网络在地下水质评价中的应用[M].上海师范大学硕士学位论文,2007,30-31
[45]K Osakada, G B Yang. Application of neural networks to an expert system for cold forging[J]. International Journal of Machine tools,2003,25(5),99-106
[46]Hagan M T, Menhaj M. Training feedforward networks with the marquardt algorithm g[J]. IEEE Transactions on Neural Networks,1994,5(6):989-993
[42]廉师友.人工智能技术导论[M].西安：西安电子科技大学出版社,2002,1-5
[43]王万森.人工智能原理及其应用[M].北京：电子工业出版社,2000,240-246
[49]吕大刚,王光远.结构智能优化设计一个新的研究方向[J].哈尔滨建筑大学学报,1999,32(4)：7-12
[50]郝永涛,赵卫东,李启炎.基于模式化知识和人工神经网络的智能设计系统框架研究[J].计算机辅助设计与图形学报,2001,13(9)：834-839
[51]Narayanan V. Systems for the prediction of process parameters[J].Journal of Materials Processing Technology,1995,54:64-69
[52]李显宏.Matlab7.x界面设计与编译技巧[M].北京：电子工业出版社,2006,203-256
[53]施晓红,周佳.精通GUI图形界面编程[M].北京：北京大学出版社,2003,164-192
[2]侯鹏.夹层橡胶基础隔震结构动力分析[J].成都：西南交通大学硕士学位论文,2003,1-2
[3]日本免震构造协会编,叶列平译.隔震结构入门.北京：科学出版社,1998,4-10
[4]李慧,杜永峰,屠锦敏,周茗如,邓学晶.叠层橡胶基础隔震技术的研究[J].甘肃工业大学学报,1999,25(2)：81-83
[5]李慧,杜永峰,屠锦敏,梁文钊,黎海南,王光照.兰州市某住宅楼工程抗震方法的优化选择[J].基建优化,2000,21(2)：30-32
[6]王社良.抗震结构设计[M].武汉：武汉理工大学出版社,2008,224-232
[7]Farzad Naeim, James M.Kelly. Design of seismic isolated structures[M]. canda, John, wiley & Sons Inc,1999,35-46
[8]GB 50011-2001,建筑抗震设计规范[S]
[9]CECS126:2001叠层橡胶支座隔震技术规程[S]
[10]日本建筑学会编,刘文光译.隔震结构设计[M].北京：地震出版社,2006,28-40
[11]张志公.大底盘多塔楼结构基础隔震性能分析与研究[M].兰州理工大学硕士学位
论文,23-30
[12]CECS160:2004建筑工程抗震性态设计通则(试用)[S]
[13]党育,杜永峰,李慧.基础隔震结构设计及施工指南[M].北京：中国水利水电出版社,2007,73-79
[14]罗四维.神经网络建造[M].北京：中国铁道出版社,1998,3-5
[15]赵林明,胡浩云,魏德华,王树谦.多层前向人工神经网络.郑州：黄河水利出版社,1999,69-90
[16]Ghaboussi.J,Joghataie.A.Active control of structures using neural networks [J]. Eng. Mec,1995,4(55):554-567
[17]魏东.非线性系统神经网络参数预测及控制[M].北京：机械工业出版社,2008,34-38
[18]张代远.神经网络新理论与方法[M].北京：清华大学出版社,2006,4-5
[19]韩立群.人工神经网络教程[M].北京：北京邮电大学出版社,2006,63-75
[20]刘伯权,王经建.基于神经网络的地震区结构经济性分析系统[J].长安大学学报,2004,24(5)：55-59
[21]黄浩,陶龙光,李幻.人工神经网络在岩土工程中的应用状况及前景[J].工业建筑,2007,37(增刊)：844-847
[22]汤皓,陈国兴.灰关联与人工神经网络在建筑物震害预测中的应用[J].地震工程与工 程振动,2006,26(3)：57-59
[23]梅青海,霍凯成.基于人工神经网络的桩基础选型研究[J].武汉理工大学学报,2004,26(12)：63-65
[24]郑浩,王全风.BP神经网络在高层结构选型中的应用[J].华侨大学学报,2003,24(1)：48-55
[25]王新刚,王起才,周岩.基于人工神经网络的大体积混凝土温度场预测[J].混凝土2006,203(9)：21-24
[26]李慧,王红丽,杜永峰.钢框架连接损伤的两种改进BP算法的比较[J].工程抗震与加固改造,2006,28(6)：39-42
[27]李慧,王红丽,杜永峰,张东兵.基于改进BP算法的刚框架连接损伤识别方法[J].甘肃科学学报,2007,19(2)：114-117
[28]杜永峰,李慧.高效神经网络训练及其在桁架损伤识别中的应用[J].华中科技大学学报,2008,25(4)：51-53
[29]杜永峰,郭剑虹.基于RBF神经网络的结构动力响应预测[J].兰州理工大学学报,2006,32(2)：111-114
[30]赵艳林,杨军平,黄剑飞,吕海波.基于BP神经网络的单层钢筋混凝土柱工业厂房震害预测[J].桂林工学院,2006,26(4)：491-496
[31]Qin Zheng,Wuzhong fu,Xiaofeng Liao,et al. A network Intrusion Detection Architecture Based on Intelligent Agents.Chong qin:computer science.2005,32(6):133-136
[32]韩立群.人工神经网络理论、设计及应用[M].北京：化学工业出版社,2002,44-56
[33]许东,吴铮.基于MATLAB 6.X的系统分析与设计-神经网络[M].西安：西安电子科技大学出版社,2002,19-24
[34]飞思科技产品研发中心.MATLAB 6.5辅助神经网络分析与设计[M].北京：电子工业出版社,2003,64-69
[35]Lee S.C,Heinbuch D.V.Training a neural-network based intrusion detector to recongnize novel attacks[J].IEEE Transactions on Systems.2001,31(4):294-299
[36]魏海坤.神经网络结构设计的理论与方法[M].北京：国防工业出版社,2007,36-39
[37]Sung A H,Mukkamala S.Identifying Important Features for Intrusion Detection Using Support Vector Machines and Neural Networks. Applications and the Internet Proceedings[J].USA:2003 symposium on Publication,2003,209-216
[38]Boozarjomehry,R.B.Svrcek,W.Y. Automatic design of neural network structures [J]. Computers and Chemical Engineering,2000,25 (7-8):1075-1088
[39]王红丽.框架结构节点损伤识别的数值与实验研究[M].兰州理工大学硕士学位论文,2007,13-27
[40]Ignizio, James P, Soltys, James R. Simultaneous design and training of ontogenic neural network classifiers[J].Computers & Operations Research,1999,23 (6):535-546
[41]何雪忠.复杂大分子体系平衡性质的神经网络预测[M].郑州：郑州大学硕士学位论文,2004,37-39
[42]王燚,郭伟.一种混沌神经网络的混沌时间序列预测[J].电子测量与仪器学报,2007,21(6)：5-9
[43]Sarkav D.Methods to speed up error BP learning aloritmn[J]. ACM Computing survey, 1995,27(3):519-592
[44]杜刚.改进的BP神经网络在地下水质评价中的应用[M].上海师范大学硕士学位论文,2007,30-31
[45]K Osakada, G B Yang. Application of neural networks to an expert system for cold forging[J]. International Journal of Machine tools,2003,25(5),99-106
[46]Hagan M T, Menhaj M. Training feedforward networks with the marquardt algorithm g[J]. IEEE Transactions on Neural Networks,1994,5(6):989-993
[42]廉师友.人工智能技术导论[M].西安：西安电子科技大学出版社,2002,1-5
[43]王万森.人工智能原理及其应用[M].北京：电子工业出版社,2000,240-246
[49]吕大刚,王光远.结构智能优化设计一个新的研究方向[J].哈尔滨建筑大学学报,1999,32(4)：7-12
[50]郝永涛,赵卫东,李启炎.基于模式化知识和人工神经网络的智能设计系统框架研究[J].计算机辅助设计与图形学报,2001,13(9)：834-839
[51]Narayanan V. Systems for the prediction of process parameters[J].Journal of Materials Processing Technology,1995,54:64-69
[52]李显宏.Matlab7.x界面设计与编译技巧[M].北京：电子工业出版社,2006,203-256
[53]施晓红,周佳.精通GUI图形界面编程[M].北京：北京大学出版社,2003,164-192