粗糙集—小波神经网络在隧道围岩分类中的应用
|
摘要
隧道工程围岩的级别是隧道围岩稳定性的尺度,施工期间的隧道围岩分类的确定是最为基础、也是最为重要的内容。本文将粗糙集、小波神经网络和围岩分类有机结合起来,对白鹤隧道围岩分类进行识别研究。结果表明:用经过粗糙集约简后的样本集作为神经网络的训练样本集,有效地简化了神经网络的结构,减少了训练步数与训练时间,并提高了网络的学习速度和判断准确率;经过粗糙集约简后的WNN判别准确率最高,识别结果更接近专家质量评价法;而BP网络判别结果与专家质量评价法相差较大;总体上,小波神经网络预判的结果要比BP神经网络预判的结果精度要高,约简后要比约简前的精度要高。
The surrounding rock grade of a tunnel is the scale of tunnel stability.The determination of rock classification is the basic and most important content in the period of tunnel excavation.In this paper,the rough set,wavelet neural networks and rock classification were combined and applied to Baihe tunnel rock classification.The training sample set of neural network is reduced by rough set,the neural network structure is effectively simplified,the training steps and the training time are decreased,and the network learning speed and accuracy is improved.The judgement accuracy by using rough set WNN is the highest,and the recognition results are the nearest to the results by expert discriminant system.The results using BP neural network has the largest difference with the expert discriminant system value.In general,the prediction accuracy using the wavelet neural networks is higher than the prediction accuracy using BP neural network,and the prediction accuracy using rough set is higher than that without rough set.
The surrounding rock grade of a tunnel is the scale of tunnel stability.The determination of rock classification is the basic and most important content in the period of tunnel excavation.In this paper,the rough set,wavelet neural networks and rock classification were combined and applied to Baihe tunnel rock classification.The training sample set of neural network is reduced by rough set,the neural network structure is effectively simplified,the training steps and the training time are decreased,and the network learning speed and accuracy is improved.The judgement accuracy by using rough set WNN is the highest,and the recognition results are the nearest to the results by expert discriminant system.The results using BP neural network has the largest difference with the expert discriminant system value.In general,the prediction accuracy using the wavelet neural networks is higher than the prediction accuracy using BP neural network,and the prediction accuracy using rough set is higher than that without rough set.
引文
[1]Deere,D.U.Technical description of rock cores forengineering purposes[J].Felsmechanik und Ingenieu-geologie,1963,1(1):16-22.
[2]Barton N.,Lien R.&Lunde J.Engineering classifica-tion of rock masses for design of the tunnel support[J].Rock Mechanics,1974,6(4):189-236.
[3]Arild Palmstrom.Characterizing rock masses by theRMi for use in practical rock engineering.Part 1:Thedevelopment of the Rock Mass index(RMi).Tunnelingand Underground Space Technology,1996,(2):175-188.
[4]Arild Palmstrom.Characterizing rock masses by theRMi for use in practical rock engineering.Part 2:Thedevelopment of the Rock Mass index(RMi).Tunnelingand Underground Space Technology,1996,(3):287-303.
[5]Bienlawski Z.T.Engineering classification of jointedmass,Trans.S.Africa Inst,Civ.Engrs,1973,15(12):335-344.
[6]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001.(Zhang Wenxiu,WuWeizhi,Liang Jiye,et al.Rough set theory and method[M].Beijing:Science Press,2001.(in Chinese))
[7]宋笑雪.粗糙集理论及其应用[J].咸阳师范学院学报,2005,20(2):29-31.(Song Xiaoxue.Rough setstheory and its application[J].Journal of Xianyang Nor-mal University,2005,20(2):29-31.(in Chinese))
[8]李建平,唐远炎.小波分析方法的应用[M].重庆:重庆大学出版社,1999.(Li Jianping,Tang Yuanyan.Application of wavelet analysis[M].Chongqing:Chongqing University Press,1999.(in Chinese))
[9]郑治真,沈萍,杨选辉,等.小波变换及其MATLAB工具的应用[M].北京:地震出版社,2002.(ZhengZhizhen,Shen Ping,Yang Xuanhui,et al.Applicationof wavelet analysis and the MATLAB tool box[M].Beijing:Earthquake Press,2002.(in Chinese))
[10]Zhang Qinghua,Benveniste A.Wavelet networks.IEEETrans.on Neural networks.1992,3(6):859-898.
[11]潘国荣.变形监测数据的小波神经网络预测方法[J].大地测量与地球动力学,2007,27(4):47-50.(Pan Guorong.Wavelet neural network predictionmethod of deformation monitoring data[J].Journal ofGeodesy and Geodynamics,2007,27(4):47-50.(inChinese))
[12]谭云亮,孙中辉,杜学东.冲击地压AE时间序列小波神经网络预测模型[J].岩石力学与工程学报,2000,19(增):1 034-1 036.(Tan Yunliang,SunZhonghui,Du Xuedong.A wavelet neural network pre-dicting model for acoustic emission time series of rockburst[J].Chinese Journal of Rock Mechanics and En-gineering,2000,19(Supp.):1 034-1 036.(in Chi-nese))
[13]孙少锐,吴继敏,魏继红.基于地质统计模型的小波神经网络在地下洞室超挖预测中的应用[J].岩石力学与工程学报,2003,22(8):1 344-1 349.(Sun Sha-orui,Wu Jimin,Wei Jihong.Application of waveletneural network based on geological statistical model in toover break prediction of underground openings[J].Chi-nese Journal of Rock Mechanics and Engineering,2003,22(8):1 344-1 349.(in Chinese))
[14]张玉祥.小波神经网络遗传算法及其在矿山压力预报中的应[J].中国有色金属学报,1999,9(2):448-452.(Zhang Yuxiang.Genetic algorithm of wavelet net-work and its application in forecasting ground pressure[J].The Chinese Journal of Nonferrous Metals,1999,9(2):448-452.(in Chinese))
[2]Barton N.,Lien R.&Lunde J.Engineering classifica-tion of rock masses for design of the tunnel support[J].Rock Mechanics,1974,6(4):189-236.
[3]Arild Palmstrom.Characterizing rock masses by theRMi for use in practical rock engineering.Part 1:Thedevelopment of the Rock Mass index(RMi).Tunnelingand Underground Space Technology,1996,(2):175-188.
[4]Arild Palmstrom.Characterizing rock masses by theRMi for use in practical rock engineering.Part 2:Thedevelopment of the Rock Mass index(RMi).Tunnelingand Underground Space Technology,1996,(3):287-303.
[5]Bienlawski Z.T.Engineering classification of jointedmass,Trans.S.Africa Inst,Civ.Engrs,1973,15(12):335-344.
[6]张文修,吴伟志,梁吉业,等.粗糙集理论与方法[M].北京:科学出版社,2001.(Zhang Wenxiu,WuWeizhi,Liang Jiye,et al.Rough set theory and method[M].Beijing:Science Press,2001.(in Chinese))
[7]宋笑雪.粗糙集理论及其应用[J].咸阳师范学院学报,2005,20(2):29-31.(Song Xiaoxue.Rough setstheory and its application[J].Journal of Xianyang Nor-mal University,2005,20(2):29-31.(in Chinese))
[8]李建平,唐远炎.小波分析方法的应用[M].重庆:重庆大学出版社,1999.(Li Jianping,Tang Yuanyan.Application of wavelet analysis[M].Chongqing:Chongqing University Press,1999.(in Chinese))
[9]郑治真,沈萍,杨选辉,等.小波变换及其MATLAB工具的应用[M].北京:地震出版社,2002.(ZhengZhizhen,Shen Ping,Yang Xuanhui,et al.Applicationof wavelet analysis and the MATLAB tool box[M].Beijing:Earthquake Press,2002.(in Chinese))
[10]Zhang Qinghua,Benveniste A.Wavelet networks.IEEETrans.on Neural networks.1992,3(6):859-898.
[11]潘国荣.变形监测数据的小波神经网络预测方法[J].大地测量与地球动力学,2007,27(4):47-50.(Pan Guorong.Wavelet neural network predictionmethod of deformation monitoring data[J].Journal ofGeodesy and Geodynamics,2007,27(4):47-50.(inChinese))
[12]谭云亮,孙中辉,杜学东.冲击地压AE时间序列小波神经网络预测模型[J].岩石力学与工程学报,2000,19(增):1 034-1 036.(Tan Yunliang,SunZhonghui,Du Xuedong.A wavelet neural network pre-dicting model for acoustic emission time series of rockburst[J].Chinese Journal of Rock Mechanics and En-gineering,2000,19(Supp.):1 034-1 036.(in Chi-nese))
[13]孙少锐,吴继敏,魏继红.基于地质统计模型的小波神经网络在地下洞室超挖预测中的应用[J].岩石力学与工程学报,2003,22(8):1 344-1 349.(Sun Sha-orui,Wu Jimin,Wei Jihong.Application of waveletneural network based on geological statistical model in toover break prediction of underground openings[J].Chi-nese Journal of Rock Mechanics and Engineering,2003,22(8):1 344-1 349.(in Chinese))
[14]张玉祥.小波神经网络遗传算法及其在矿山压力预报中的应[J].中国有色金属学报,1999,9(2):448-452.(Zhang Yuxiang.Genetic algorithm of wavelet net-work and its application in forecasting ground pressure[J].The Chinese Journal of Nonferrous Metals,1999,9(2):448-452.(in Chinese))
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心