基于SVC参数优化的地震次生地质灾害危险性评价
|
摘要
针对地震次生地质灾害危险性评价影响因素的复杂性和多变性的特点,提出了基于GA、PSO和K-CV三种优化支持向量分类机参数的地震次生地质灾害危险性评价方法.该方法既利用了支持向量分类机求解速度快、易于描述非线性关系的优良特性,同时利用了GA、K-CV和PSO算法快速优化的特点,可实现支持向量分类机模型参数的自动化优选,具有收敛速度快、精度高的特点.将该模型用于地震次生地质灾害危险性评价,计算结果验证了该方法的有效性。
According to the characteristics of the complexity and variability of factors,a risk degree of earthquake secondary geological hazards evaluation method was proposed based on GA、PSO and K-CV optimized parameters of support vector classification.The method not only has the excellent characteristics of fast solving speed and describing nonlinear relationship of SVC easily,but also has the characteristic of fast optimization.GA、PSO and K-CV were used to search the optimum parameters of SVC,then the optimal model was applied to evaluate the risk degree of earthquake secondary geological hazards.The results show that the method is reasonable and feasible.
According to the characteristics of the complexity and variability of factors,a risk degree of earthquake secondary geological hazards evaluation method was proposed based on GA、PSO and K-CV optimized parameters of support vector classification.The method not only has the excellent characteristics of fast solving speed and describing nonlinear relationship of SVC easily,but also has the characteristic of fast optimization.GA、PSO and K-CV were used to search the optimum parameters of SVC,then the optimal model was applied to evaluate the risk degree of earthquake secondary geological hazards.The results show that the method is reasonable and feasible.
引文
[1]VAPNIK V.Statistical learning theory[M].New York:John Wiley,1998.
[2]XU P,CHAN A K.Support vector machine for multi-classsignal classification with unbalanced samples[C]∥Proceedings of the International Joint Conference on NeuralNetworks 2003.Portland:IEEE,2003:1116-1119.
[3]郭小东,田杰,王威,等.基于GA-SVR的建筑物不均匀液化震陷预测方法研究[J].北京工业大学学报,2011,37(6):829-835.GUO Xiao-dong,TIAN Jie,WANG Wei,et al.Methodfor building settlements prediction due to earthquakeliquefaction based on GA-SVR[J].Journal of BeijingUniversity of Technology,2011,37(6):829-835.(inChinese)
[4]MATLAB中文论坛.MATLAB神经网络30个案例分析[M].北京:北京航空航天大学出版社,2010.
[5]倪丽萍,倪志伟,李锋刚,等.基于蚁群算法的SVM模型选择研究[J].计算机技术与发展,2007,17(9):95-98.NI Li-ping,NI Zhi-wei,LI Feng-gang,et al.SVM modelselection based on ant colony algorithm[J].ComputerTechnology and Development,2007,17(9):95-98.(inChinese)
[6]王威.基于复杂性理论的城市抗震防灾规划相关评价方法研究[D].北京:北京工业大学建筑工程学院,2010.WANG Wei.Research on relevant evaluation methods forurban planning on earthquake resistance and hazardousprevention based on complexity theory[D].Beijing:Institute of Civil Engineering of Beijing University ofTechnology,2010.(in Chinese)
[7]刘勇健.基于聚类-二叉树支持向量机的砂土液化预测模型[J].岩土力学,2008,39(10):2764-2768.LIU Yong-jian.Support vector machine model forpredicting sand liquefaction based on clustering binary treealgorithm[J].Rock and Soil Mechanics,2008,39(10):2764-2768.(in Chinese)
[8]刘丽,王士革.滑坡、泥石流区域危险度二级模糊综合评判初探[J].自然灾害学报,1996,5(3):51-59.LIU Li,WANG Shi-ge.Preliminary research of two levelfuzzy comprehensive evaluation on landslide and debrisflow risk degree of a district[J].Journal of NaturalDisasters,1996,5(3):51-59.(in Chinese)
[9]刘丽,王士革.云南昭通滑坡泥石流危险度模糊综合评判[J].山地研究,1995,13(4):261-269.LIU Li,WANG Shi-ge.Fuzzy comprehensive evaluationon landslide and debris flow risk degree in Zhaotong,Yunnan[J].Mountain Research,1995,13(4):261-269.(in Chinese)
[10]王威,田杰,马东辉,等.基于分形维数权重的泥石流危险度评价[J].山地学报,2011,29(6):747-752.WANG Wei,TIAN Jie,MA Dong-hui,et al.Evaluationof dangerous degree of debris flow based on fractaldimension weight[J].Journal of Mountain Science,2011,29(6):747-752.(in Chinese)
[11]高学金,王普,齐咏生,等.LS-SVM和SVM在发酵过程建模中的比较[J].北京工业大学学报,2010,36(1):7-12.GAO Xue-jin,WANG Pu,QI Yong-sheng,et al.Comparison studies of LS-SVM and SVM on modeling forfermentation process[J].Journal of Beijing University ofTechnology,2010,36(1):7-12.(in Chinese)
[12]杜京义,侯媛彬.基于遗传算法的支持向量回归机参数选取[J].系统工程与电子技术,2006,28(9):1430-1433.DU Jing-yi,HOU Yuan-bin.Parameters selection ofsupport vector regression by genetic algorithms[J].Systems Engineering and Electronics,2006,28(9):1430-1433.(in Chinese)
[13]王威,苏经宇,马东辉,等.城市综合承灾能力评价的粒子群优化投影寻踪模型[J].北京工业大学学报,2012,38(8):1174-1179.WANG Wei,SU Jing-yu,MA Dong-hui,et al.Evaluation method of urban comprehensive disaster-carrying capability based on PSO-PPE[J].Journal ofBeijing University of Technology,2012,38(8):1174-1179.(in Chinese)
[2]XU P,CHAN A K.Support vector machine for multi-classsignal classification with unbalanced samples[C]∥Proceedings of the International Joint Conference on NeuralNetworks 2003.Portland:IEEE,2003:1116-1119.
[3]郭小东,田杰,王威,等.基于GA-SVR的建筑物不均匀液化震陷预测方法研究[J].北京工业大学学报,2011,37(6):829-835.GUO Xiao-dong,TIAN Jie,WANG Wei,et al.Methodfor building settlements prediction due to earthquakeliquefaction based on GA-SVR[J].Journal of BeijingUniversity of Technology,2011,37(6):829-835.(inChinese)
[4]MATLAB中文论坛.MATLAB神经网络30个案例分析[M].北京:北京航空航天大学出版社,2010.
[5]倪丽萍,倪志伟,李锋刚,等.基于蚁群算法的SVM模型选择研究[J].计算机技术与发展,2007,17(9):95-98.NI Li-ping,NI Zhi-wei,LI Feng-gang,et al.SVM modelselection based on ant colony algorithm[J].ComputerTechnology and Development,2007,17(9):95-98.(inChinese)
[6]王威.基于复杂性理论的城市抗震防灾规划相关评价方法研究[D].北京:北京工业大学建筑工程学院,2010.WANG Wei.Research on relevant evaluation methods forurban planning on earthquake resistance and hazardousprevention based on complexity theory[D].Beijing:Institute of Civil Engineering of Beijing University ofTechnology,2010.(in Chinese)
[7]刘勇健.基于聚类-二叉树支持向量机的砂土液化预测模型[J].岩土力学,2008,39(10):2764-2768.LIU Yong-jian.Support vector machine model forpredicting sand liquefaction based on clustering binary treealgorithm[J].Rock and Soil Mechanics,2008,39(10):2764-2768.(in Chinese)
[8]刘丽,王士革.滑坡、泥石流区域危险度二级模糊综合评判初探[J].自然灾害学报,1996,5(3):51-59.LIU Li,WANG Shi-ge.Preliminary research of two levelfuzzy comprehensive evaluation on landslide and debrisflow risk degree of a district[J].Journal of NaturalDisasters,1996,5(3):51-59.(in Chinese)
[9]刘丽,王士革.云南昭通滑坡泥石流危险度模糊综合评判[J].山地研究,1995,13(4):261-269.LIU Li,WANG Shi-ge.Fuzzy comprehensive evaluationon landslide and debris flow risk degree in Zhaotong,Yunnan[J].Mountain Research,1995,13(4):261-269.(in Chinese)
[10]王威,田杰,马东辉,等.基于分形维数权重的泥石流危险度评价[J].山地学报,2011,29(6):747-752.WANG Wei,TIAN Jie,MA Dong-hui,et al.Evaluationof dangerous degree of debris flow based on fractaldimension weight[J].Journal of Mountain Science,2011,29(6):747-752.(in Chinese)
[11]高学金,王普,齐咏生,等.LS-SVM和SVM在发酵过程建模中的比较[J].北京工业大学学报,2010,36(1):7-12.GAO Xue-jin,WANG Pu,QI Yong-sheng,et al.Comparison studies of LS-SVM and SVM on modeling forfermentation process[J].Journal of Beijing University ofTechnology,2010,36(1):7-12.(in Chinese)
[12]杜京义,侯媛彬.基于遗传算法的支持向量回归机参数选取[J].系统工程与电子技术,2006,28(9):1430-1433.DU Jing-yi,HOU Yuan-bin.Parameters selection ofsupport vector regression by genetic algorithms[J].Systems Engineering and Electronics,2006,28(9):1430-1433.(in Chinese)
[13]王威,苏经宇,马东辉,等.城市综合承灾能力评价的粒子群优化投影寻踪模型[J].北京工业大学学报,2012,38(8):1174-1179.WANG Wei,SU Jing-yu,MA Dong-hui,et al.Evaluation method of urban comprehensive disaster-carrying capability based on PSO-PPE[J].Journal ofBeijing University of Technology,2012,38(8):1174-1179.(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