基于遗传优化神经网络的边坡稳定性评价
|
摘要
由于引起滑坡的因素复杂,传统预测方法难以得到高精度的结果。文中利用遗传算法(GA)全局搜索能力强、不易陷入局部极小值的特点对样本的初始权值和阈值进行优化处理,使得前馈型神经网络(BP)在学习和预测时能够得到一个最佳的权值和阈值,从而探索出影响滑坡的因子与边坡稳定性之间潜在的关系。从仿真结果可知:优化权值后的BP神经网络得到边坡稳定性的判对率达到100%,而随机权值BP神经网络的判对率仅为54.5%,判对率提高了45.5%;安全系数较随机权值BP神经网络的平均误差提高了6.08%。因此,优化BP神经网络的预测精度得到明显提高,在今后边坡稳定性的实际应用评价中可作为一种有效的辅助手段。
The conventional prediction method is difficult to obtain the highly-accurate prediction results due to the complex factor of landslides. The global search ability and local minimum avoidance characteristic of genetic algorithm(GA)are used to optimize the initial weight and threshold of the sample,which makes BP neural network for learning and forecasting get the optimal weight and threshold,and can explore the potential relationship between the factors affecting the landslide and slope stability. The simulation results show that the correct judgement accuracy of slope stability obtained by BP neural network after weight optimization can reach up to 100%,which is increased by 45.5% than that of 54.5% obtained by random weigh BP neural network;the average error of the safety factor obtained by BP neural network after weight optimization is increased by6.08% than that of the random weight BP neural network. The prediction accuracy of BP neural network is improved obviously,which acts as an effective auxiliary method in the practical application evaluation of slope stability in the future.
The conventional prediction method is difficult to obtain the highly-accurate prediction results due to the complex factor of landslides. The global search ability and local minimum avoidance characteristic of genetic algorithm(GA)are used to optimize the initial weight and threshold of the sample,which makes BP neural network for learning and forecasting get the optimal weight and threshold,and can explore the potential relationship between the factors affecting the landslide and slope stability. The simulation results show that the correct judgement accuracy of slope stability obtained by BP neural network after weight optimization can reach up to 100%,which is increased by 45.5% than that of 54.5% obtained by random weigh BP neural network;the average error of the safety factor obtained by BP neural network after weight optimization is increased by6.08% than that of the random weight BP neural network. The prediction accuracy of BP neural network is improved obviously,which acts as an effective auxiliary method in the practical application evaluation of slope stability in the future.
引文
[1]赵亚军,张恩龙,巩志忠.BP神经网络在边坡稳定性预测中的应用[J].西部探矿工程,2014,26(2):23-25.ZHAO Y J,ZHANG E L,GONG Z Z. Application of BP neural network in slope stability prediction[J]. West-China exploration engineering,2014,26(2):23-25.
[2] SHANGGUAN Z C,LI S,LUAN M. Intelligent forecasting method for slope stability estimation by using probabilistic neural networks[J]. Electronic journal of geotechnical engineering,2009(13):1-10.
[3]胡添翼,戴波,何启,等.基于随机森林分类算法的边坡稳定预测模型[J].人民黄河,2017,39(5):115-118.HU T Y,DAI B,HE Q,et al. Slope stability prediction model based on random forest classification algorithm[J]. Yellow river,2017,39(5):115-118.
[4]邹义怀,江成玉,李春辉.人工神经网络在边坡稳定性预测中的应用[J].矿冶,2011,20(4):38-41.ZOU Y H,JIANG C Y,LI C H. Application of artificial neu-ral network in slope stability prediction[J]. Mining and metal-lurgy,2011,20(4):38-41.
[5]赵胜利,吴雅琴,刘燕,等.基于SOM-BP复合神经网络的边坡稳定性分析[J].河北农业大学学报,2007,30(3):105-108.ZHAO S L,WU Y Q,LIU Y,et al. Analysis of slop stabilitybased on SOM-BP neural network[J]. Journal of AgriculturalUniversity of Hebei,2007,30(3):105-108.
[6]张飞,段志峰.RBF神经网络在边坡稳定性中的应用研究[J].西部探矿工程,2011,23(8):20-21.ZHANG F,DUAN Z F. Application research on RBF neuralnetwork in slope stability[J]. West-China exploration enginee-ring,2011,23(8):20-21.
[7] SHU H E,WANG J D,WANG H,et al. The evaluation ofloess slope stability based on combination of information diffu-sion theory and BP neural network[J]. Journal of NorthwestUniversity,2008,38(6):983-988.
[8]向超文,徐锦洪,李焜,等.人工神经网络边坡稳定预报模型[J].苏州科技学院学报(工程技术版),2006,19(2):1-9.XIANG C W,XU J H,LI K,et al. Prediction of the slope sta-bility on artificial neural network[J]. Journal of University ofScience and Technology of Suzhou(engineering and technolo-gy),2006,19(2):1-9.
[9]汪茜,李广杰,郑百功,等.自适应BP神经网络在边坡稳定性预测中的应用[J].人民黄河,2010,32(4):120-121.WANG Q,LI G J,ZHENG B G,et al. Application of adap-tive BP neural network in slop stability prediction[J]. Yellowriver,2010,32(4):120-121.
[10]赵鹏,王斐,刘慧婷,等.基于深度学习的手绘草图识别[J].四川大学学报(工程科学版),2016,48(3):94-99.ZHAO P,WANG F,LIU H T,et al. Sketch recognition usingdeep learning[J]. Journal of Sichuan University(engineeringscience edition),2016,48(3):94-99.
[11]章毅,郭泉,王建勇.大数据分析的神经网络方法[J].四川大学学报(工程科学版),2017,49(1):9-18.ZHANG Y,GUO Q,WANG J Y. Big data analysis usingneural network[J]. Journal of Sichuan University(engineeringscience edition),2017,49(1):9-18.
[12]向良成,肖利洪,李梅,等.基于反向传播神经网络的前列腺癌诊断系统的诊断价值[J].四川大学学报(医学版),2016,47(1):77-80.XIANG L C,XIAO L H,LI M,et al. Diagnosis values ofback propagation neural network integrating age,transrectalultrasound characteristics and serum PSA for prostate cancer[J]. Journal of Sichuan University(medical science edition),2016,47(1):77-80.
[13]王静,杜勇,赵忠华.神经网络算法在特色农产品品质分类中的应用[J].四川大学学报(自然科学版),2016,53(4):805-808.WANG J,DU Y,ZHAO Z H. The application of neural net-work algorithm in agricultural product quality classification[J]. Journal of Sichuan University(natural science edition),2016,53(4):805-808.
[14] KAUNDA R B,CHASE R B,KEHEW A E,et al. Neuralnetwork modeling applications in active slope stability problems[J]. Environmental earth sciences,2010,60(7):1545-1558.
[15] HUANG Z,CUI J,LIU H. Chaotic neural network methodfor slope stability prediction[J]. Chinese journal of rock me-chanics&engineering,2004,23(22):3808-3812.
[16]张玲,王玲,吴桐.基于改进的粒子群算法优化反向传播神经网络的热舒适度预测模型[J].计算机应用,2014,34(3):775-779.ZHANG L,WANG L,WU T. Thermal comfort predictionmodel based on improved particle swarm optimization-backpropagation neural network[J]. Computer applications,2014,34(3):775-779.
[17] ZHANG X L,JUN H U,ZHAO T Y. Geometric parametersoptimization of dump slope based on BP neural network-geneticalgorithm[J]. Mining&metallurgical engineering,2017(2):1-4.
[18]于涛.BP网络自适应学习率算法分析[D].大连:大连理工大学,2011.YU Tao. Analysis of BP network adaptive learning rate algo-rithm[D]. Dalian:Dalian University of Technology,2011.
[19]莫秋金,莫显德,万谦.改进BP神经网络在边坡稳定性分析中的运用[J].西部交通科技,2017(4):16-18.MO Q J,MO X D,WAN Q. Application of improved BP neu-ral network in slope stability analysis[J]. Western China com-munication science&technology,2017(4):16-18.
[20] ALVANITOPOULOS P F,ANDREADIS I,ELENAS A. A ge-netic algorithm for the classification of earthquake damages inbuildings[C]//Proceedings of 2009 Artificial Intelligence Ap-plications and Innovations Conference on Artificial Intelli-gence Applications and Innovations. Boston:Springer,2009:341-346.
[21] MAULIK U,BANDYOPADHYAY S. Genetic algorithm-basedclustering technique[J]. Pattern recognition,2004,33(9):1455-1465.
[22] GOLDBERG D E. Genetic algorithm in search,optimization,and machine learning[M]. MA:Addison-Wesley,1989.
[23]卢才金,胡厚田,徐建平,等.改进的BP网络在岩质边坡稳定性评判中的应用[J].岩石力学与工程学报,1999,18(3):303-307.LU C J,HU H T,XU J P,et al. Application of improvedback-propagation network in the evaluation of railway rockslope[J]. Chinese journal of rock mechanics and engineering,1999,18(3):303-307.
[24]冯夏庭,王泳嘉.边坡稳定性的神经网络估计[J].工程地质学报,1995,3(4):54-61.FENG X T,WANG Y J. Neural network estimation of slopestability[J]. Journal of engineering geology,1995,3(4):54-61.
[2] SHANGGUAN Z C,LI S,LUAN M. Intelligent forecasting method for slope stability estimation by using probabilistic neural networks[J]. Electronic journal of geotechnical engineering,2009(13):1-10.
[3]胡添翼,戴波,何启,等.基于随机森林分类算法的边坡稳定预测模型[J].人民黄河,2017,39(5):115-118.HU T Y,DAI B,HE Q,et al. Slope stability prediction model based on random forest classification algorithm[J]. Yellow river,2017,39(5):115-118.
[4]邹义怀,江成玉,李春辉.人工神经网络在边坡稳定性预测中的应用[J].矿冶,2011,20(4):38-41.ZOU Y H,JIANG C Y,LI C H. Application of artificial neu-ral network in slope stability prediction[J]. Mining and metal-lurgy,2011,20(4):38-41.
[5]赵胜利,吴雅琴,刘燕,等.基于SOM-BP复合神经网络的边坡稳定性分析[J].河北农业大学学报,2007,30(3):105-108.ZHAO S L,WU Y Q,LIU Y,et al. Analysis of slop stabilitybased on SOM-BP neural network[J]. Journal of AgriculturalUniversity of Hebei,2007,30(3):105-108.
[6]张飞,段志峰.RBF神经网络在边坡稳定性中的应用研究[J].西部探矿工程,2011,23(8):20-21.ZHANG F,DUAN Z F. Application research on RBF neuralnetwork in slope stability[J]. West-China exploration enginee-ring,2011,23(8):20-21.
[7] SHU H E,WANG J D,WANG H,et al. The evaluation ofloess slope stability based on combination of information diffu-sion theory and BP neural network[J]. Journal of NorthwestUniversity,2008,38(6):983-988.
[8]向超文,徐锦洪,李焜,等.人工神经网络边坡稳定预报模型[J].苏州科技学院学报(工程技术版),2006,19(2):1-9.XIANG C W,XU J H,LI K,et al. Prediction of the slope sta-bility on artificial neural network[J]. Journal of University ofScience and Technology of Suzhou(engineering and technolo-gy),2006,19(2):1-9.
[9]汪茜,李广杰,郑百功,等.自适应BP神经网络在边坡稳定性预测中的应用[J].人民黄河,2010,32(4):120-121.WANG Q,LI G J,ZHENG B G,et al. Application of adap-tive BP neural network in slop stability prediction[J]. Yellowriver,2010,32(4):120-121.
[10]赵鹏,王斐,刘慧婷,等.基于深度学习的手绘草图识别[J].四川大学学报(工程科学版),2016,48(3):94-99.ZHAO P,WANG F,LIU H T,et al. Sketch recognition usingdeep learning[J]. Journal of Sichuan University(engineeringscience edition),2016,48(3):94-99.
[11]章毅,郭泉,王建勇.大数据分析的神经网络方法[J].四川大学学报(工程科学版),2017,49(1):9-18.ZHANG Y,GUO Q,WANG J Y. Big data analysis usingneural network[J]. Journal of Sichuan University(engineeringscience edition),2017,49(1):9-18.
[12]向良成,肖利洪,李梅,等.基于反向传播神经网络的前列腺癌诊断系统的诊断价值[J].四川大学学报(医学版),2016,47(1):77-80.XIANG L C,XIAO L H,LI M,et al. Diagnosis values ofback propagation neural network integrating age,transrectalultrasound characteristics and serum PSA for prostate cancer[J]. Journal of Sichuan University(medical science edition),2016,47(1):77-80.
[13]王静,杜勇,赵忠华.神经网络算法在特色农产品品质分类中的应用[J].四川大学学报(自然科学版),2016,53(4):805-808.WANG J,DU Y,ZHAO Z H. The application of neural net-work algorithm in agricultural product quality classification[J]. Journal of Sichuan University(natural science edition),2016,53(4):805-808.
[14] KAUNDA R B,CHASE R B,KEHEW A E,et al. Neuralnetwork modeling applications in active slope stability problems[J]. Environmental earth sciences,2010,60(7):1545-1558.
[15] HUANG Z,CUI J,LIU H. Chaotic neural network methodfor slope stability prediction[J]. Chinese journal of rock me-chanics&engineering,2004,23(22):3808-3812.
[16]张玲,王玲,吴桐.基于改进的粒子群算法优化反向传播神经网络的热舒适度预测模型[J].计算机应用,2014,34(3):775-779.ZHANG L,WANG L,WU T. Thermal comfort predictionmodel based on improved particle swarm optimization-backpropagation neural network[J]. Computer applications,2014,34(3):775-779.
[17] ZHANG X L,JUN H U,ZHAO T Y. Geometric parametersoptimization of dump slope based on BP neural network-geneticalgorithm[J]. Mining&metallurgical engineering,2017(2):1-4.
[18]于涛.BP网络自适应学习率算法分析[D].大连:大连理工大学,2011.YU Tao. Analysis of BP network adaptive learning rate algo-rithm[D]. Dalian:Dalian University of Technology,2011.
[19]莫秋金,莫显德,万谦.改进BP神经网络在边坡稳定性分析中的运用[J].西部交通科技,2017(4):16-18.MO Q J,MO X D,WAN Q. Application of improved BP neu-ral network in slope stability analysis[J]. Western China com-munication science&technology,2017(4):16-18.
[20] ALVANITOPOULOS P F,ANDREADIS I,ELENAS A. A ge-netic algorithm for the classification of earthquake damages inbuildings[C]//Proceedings of 2009 Artificial Intelligence Ap-plications and Innovations Conference on Artificial Intelli-gence Applications and Innovations. Boston:Springer,2009:341-346.
[21] MAULIK U,BANDYOPADHYAY S. Genetic algorithm-basedclustering technique[J]. Pattern recognition,2004,33(9):1455-1465.
[22] GOLDBERG D E. Genetic algorithm in search,optimization,and machine learning[M]. MA:Addison-Wesley,1989.
[23]卢才金,胡厚田,徐建平,等.改进的BP网络在岩质边坡稳定性评判中的应用[J].岩石力学与工程学报,1999,18(3):303-307.LU C J,HU H T,XU J P,et al. Application of improvedback-propagation network in the evaluation of railway rockslope[J]. Chinese journal of rock mechanics and engineering,1999,18(3):303-307.
[24]冯夏庭,王泳嘉.边坡稳定性的神经网络估计[J].工程地质学报,1995,3(4):54-61.FENG X T,WANG Y J. Neural network estimation of slopestability[J]. Journal of engineering geology,1995,3(4):54-61.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |