基于自适应差分演化的特征选择算法在石油储层识别中的应用
|
摘要
近年来我国石油产量跟不上需求,供需矛盾进一步凸显,导致石油的对外依存度已经连续几年超过警戒线,为了缓解供需矛盾,石油的增储上产是一种有效措施,但精确地识别石油储层成为增储上产的一大难题,而特征选择是精确识别石油储层的有效保障.本文提出了一种增强型自适应差分演化算法,即ESADE算法,在算法中使用了双种群的概念,构造了一个简单的双层差分演化,并且在算法的选择操作中加入模拟退火的思想;接着将ESADE算法作为特征选择的搜索策略,将ReliefF算法、BIF算法、FCBF算法及随机抽选特征算法作为评价准则库,SOM神经网络算法、模糊C均值算法、K均值算法和K近邻算法作为分类器库,得到了一种基于ESADE的特征选择算法.然后将此算法应用于某油田oil81、oil82、oil83、oil84和oil85五口井的测井数据集上进行石油储层的油层、差油层、水层和干层的分类识别,并与未进行特征选择直接进行分类的结果进行比较及相同分类正确率下不同分类算法组合及不同属性选择的比较.实验结果表明与SOM神经网络算法、模糊C均值算法、K均值算法及K近邻算法这四种分类算法相比,基于ESADE的特征选择算法能在利用较少属性的同时提高分类准确率,并能够提供不同的属性和分类算法的最优组合方案.
Recent years,a significant oil shortage is witnessed in China,making the contradiction between oil supply and demand prominent.Consequently,the dependency on oil import becomes severe.As effective measures,the growth in oil reserve and the improvement in oil output can make great contributions in easing the tensions in oil consumption.However,there are difficulties in identifying the oil reservoirs accurately.As feature selection can be used as an effective means in solving the problem,this paper proposes an enhanced self-adaptive differential evolution algorithm,namely ESADE algorithm to cope with the situation.In this algorithm,it uses dual populations and constructs a simple double-layer differential evolution.At the same time,its selection operation is combined with simulated annealing algorithm.And,with ESADE as the search strategy,ReliefF,BIF,FCBF and randomly selected algorithm as the pool of evaluation standard,SOM neural network,fuzzy c-means,k-means and k-nearest neighbor algorithm as the pool of classifier,the paper builds a feature selection algorithm based on ESADE.Then,it applies the algorithm in the classification of oil layer,poor oil layer,water layer and dry layer based on the datasets from five wells namely oil81,oil82,oil83,oil84,oil85.After that,the accuracy of the classification under this method is compared to that under direct classification and the situation when accuracy is the same while algorithm combination and logging attributes differ.The result reflects that,compared to SOM neural network algorithm,fuzzy C-means algorithm,K-means algorithm and the K-nearest neighbor algorithm,feature selection algorithm method based on ESADE can ensure better accuracy in classification when attributes are relatively less.Besides,this algorithm can provide the optimal combination of different attributes and classification algorithms.
Recent years,a significant oil shortage is witnessed in China,making the contradiction between oil supply and demand prominent.Consequently,the dependency on oil import becomes severe.As effective measures,the growth in oil reserve and the improvement in oil output can make great contributions in easing the tensions in oil consumption.However,there are difficulties in identifying the oil reservoirs accurately.As feature selection can be used as an effective means in solving the problem,this paper proposes an enhanced self-adaptive differential evolution algorithm,namely ESADE algorithm to cope with the situation.In this algorithm,it uses dual populations and constructs a simple double-layer differential evolution.At the same time,its selection operation is combined with simulated annealing algorithm.And,with ESADE as the search strategy,ReliefF,BIF,FCBF and randomly selected algorithm as the pool of evaluation standard,SOM neural network,fuzzy c-means,k-means and k-nearest neighbor algorithm as the pool of classifier,the paper builds a feature selection algorithm based on ESADE.Then,it applies the algorithm in the classification of oil layer,poor oil layer,water layer and dry layer based on the datasets from five wells namely oil81,oil82,oil83,oil84,oil85.After that,the accuracy of the classification under this method is compared to that under direct classification and the situation when accuracy is the same while algorithm combination and logging attributes differ.The result reflects that,compared to SOM neural network algorithm,fuzzy C-means algorithm,K-means algorithm and the K-nearest neighbor algorithm,feature selection algorithm method based on ESADE can ensure better accuracy in classification when attributes are relatively less.Besides,this algorithm can provide the optimal combination of different attributes and classification algorithms.
引文
[1]边肇祺,张学工.模式识别[M].2版.北京:清华大学出版社,2001:176-210.Bian Zhaoqi,Zhang Xuegong.Pattern recognition[M].2nd ed.Beijing:Tsinghua University Press,2001:176-210.
[2]Dash M,Liu H.Feature selection for classification[J].Intelligent Data Analysis,1977,1(3):131-156.
[3]Kira K,Rendell L A.The feature selection problem:Traditional methods and a new algorithm[C]//Proceedings of 9th National Conference on AI,1992:129-134.
[4]Kononenko I.Estimating attributes:Analysis and extension of relief[C]//Proceedings of European Conference on Machine Learning,1994:171-182.
[5]Jain A K,Robert P W,Mao J C.Statistical pattern recognition:A review[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(1):4-37.
[6]Yu L,Liu H.Efficient feature selection via analysis of relevance and redundancy[J].Journal of Machine Learning Research,2004,5(1):1205-1224.
[7]Guyon I,Weston J,Barnhill S,et al.Gene selection for cancer classification using support vector machines[J].Machine Learning,2002,46:389-422.
[8]Shieh M D,Yang C C.Multiclass SVM-RFE for product form feature selection[J].Expert Systems with Applications,2008,35(1):531-541.
[9]Bacauskiene M,Verikas A,Gelzinis A,et al.A feature selection technique for generation of classification committees and its application to categorization of laryngeal images[J].Pattern Recognition,2009,42(5):645-654.
[10]张永波,游录金,陈杰新.基于模拟退火的多标记数据特征选择[J].计算机工程与设计,2011,32(7):2494-2496.Zhang Yongbo,You Lujin,Chen Jiexin.Feature selection for multi-lable data by using simulated annealing[J].Computer Engineering and Design,2011,32(7):2494-2496.
[11]戚孝铭,施亮.基于模拟退火及蜂群算法的优化特征选择算法[J].计算机工程与设计,2013,34(8):2917-2921.Qi Xiaoming,Shi Liang.Improved feature selection algorithm based on simulated annealing algorithm and artificial bee colony algorithm[J].Computer Engineering and Design,2013,34(8):2917-2921.
[12]邬开俊,鲁怀伟.采用并行协同进化遗传算法的文本特征选择[J].系统工程理论与实践,2012,32(10):2215-2220.Wu Kaijun,Lu Huaiwei.PCEGA used to solve text feature selection[J].Systems Engineering—Theory&Practice,2012,32(10):2215-2220.
[13]戴大蒙,慕德俊.非完备信息系统的启发式特征选择遗传算法[J].电子学报,2013,41(3):451-455.Dai Dameng,Mu Dejun.Heuristic genetic algorithm for feature selection in incomplete information systems[J].Acta Electronica Sinica,2013,41(3):451-455.
[14]吴克寿,陈玉明,谢荣生,等.基于粗糙集与蚁群优化算法的特征选择方法研究[J].计算机应用研究,2011,28(7):2436-2438.Wu Keshou,Chen Yuming,Xie Rongsheng,et al.Rough sets and ant colony optimization based feature selection[J].Application Research of Computers,2011,28(7):2436-2438.
[15]杨鸿章.基于蚁群算法特征选择的语音情感识别[J].计算机仿真,2013(4):377-381.Yang Hongzhang.Feature selection of speech emotional recognition based on ant colony optimization algorithm[J].Computer Simulation,2013(4):377-381.
[16]尹宏鹏,刘兆栋,罗显科,等.一种基于粒子群优化的目标跟踪特征选择算法[J].计算机工程与应用,2013,49(17):164-168.Yin Hongpeng,Liu Zhaodong,Luo Xianke,et al.Target tracking feature selection algorithm based on particle swarm optimization[J].Computer Engineering and Applications,2013,49(17):164-168.
[17]姚旭,王晓丹,张玉玺,等.基于自适应t分布变异的粒子群特征选择方法[J].系统工程与电子技术,2013,35(6):1335-1341.Yao Xu,Wang Xiaodan,Zhang Yuxi,et al.Feature selection algorithm using PSO with adaptive mutation based t distribution[J].Journal of Systems Engineering and Electronics,2013,35(6):1335-1341.
[18]Storn R,Price K V.Minimizing the real functions of the ICEC 1996 contest by differential evolution[C]//Proceedings of IEEE International Conference Evolutionary Computer,1996:842-844.
[19]尹小娟,肖勤,琚报德.基于SOM的三维人脸表情聚类[J].信息系统工程,2013(3):137.
[20]孟海东,马娜娜,宋宇辰,等.基于密度函数加权的模糊C均值聚类算法研究[J].计算机工程与应用,2012,48(27):123-127.Meng Haidong,Ma Nana,Song Yuchen,et al.Research on fuzzy C-means clustering algorithm based on density function weighted[J].Computer Engineering and Applications,2012,48(27):123-127.
[21]郭均鹏,陈颖,李汶华.一般分布区间型符号数据的K均值聚类方法[J].管理科学学报,2013,16(3):21-28.Guo Junpeng,Chen Ying,Li Wenhua.K-means clustering of generally distributed interval symbolic data[J].Journal of Management Sciences in China,2013,16(3):21-28.
[22]潘锋,王建东,顾其威,等.基于图的特征选择算法[J].计算机工程,2012,38(9):197-198.Pan Feng,Wang Jiandong,Gu Qiwei,et al.Feature selection algorithm based on graph[J].Computer Engineering,2012,38(9):197-198.
[23]Liang J J,Suganthan P N,Deb K.Novel composition test functions for numerical global optimizationfC]//IEEE Swarm Intelligence Symposium 2005,Pasadena,California,2005:68-75.
[2]Dash M,Liu H.Feature selection for classification[J].Intelligent Data Analysis,1977,1(3):131-156.
[3]Kira K,Rendell L A.The feature selection problem:Traditional methods and a new algorithm[C]//Proceedings of 9th National Conference on AI,1992:129-134.
[4]Kononenko I.Estimating attributes:Analysis and extension of relief[C]//Proceedings of European Conference on Machine Learning,1994:171-182.
[5]Jain A K,Robert P W,Mao J C.Statistical pattern recognition:A review[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(1):4-37.
[6]Yu L,Liu H.Efficient feature selection via analysis of relevance and redundancy[J].Journal of Machine Learning Research,2004,5(1):1205-1224.
[7]Guyon I,Weston J,Barnhill S,et al.Gene selection for cancer classification using support vector machines[J].Machine Learning,2002,46:389-422.
[8]Shieh M D,Yang C C.Multiclass SVM-RFE for product form feature selection[J].Expert Systems with Applications,2008,35(1):531-541.
[9]Bacauskiene M,Verikas A,Gelzinis A,et al.A feature selection technique for generation of classification committees and its application to categorization of laryngeal images[J].Pattern Recognition,2009,42(5):645-654.
[10]张永波,游录金,陈杰新.基于模拟退火的多标记数据特征选择[J].计算机工程与设计,2011,32(7):2494-2496.Zhang Yongbo,You Lujin,Chen Jiexin.Feature selection for multi-lable data by using simulated annealing[J].Computer Engineering and Design,2011,32(7):2494-2496.
[11]戚孝铭,施亮.基于模拟退火及蜂群算法的优化特征选择算法[J].计算机工程与设计,2013,34(8):2917-2921.Qi Xiaoming,Shi Liang.Improved feature selection algorithm based on simulated annealing algorithm and artificial bee colony algorithm[J].Computer Engineering and Design,2013,34(8):2917-2921.
[12]邬开俊,鲁怀伟.采用并行协同进化遗传算法的文本特征选择[J].系统工程理论与实践,2012,32(10):2215-2220.Wu Kaijun,Lu Huaiwei.PCEGA used to solve text feature selection[J].Systems Engineering—Theory&Practice,2012,32(10):2215-2220.
[13]戴大蒙,慕德俊.非完备信息系统的启发式特征选择遗传算法[J].电子学报,2013,41(3):451-455.Dai Dameng,Mu Dejun.Heuristic genetic algorithm for feature selection in incomplete information systems[J].Acta Electronica Sinica,2013,41(3):451-455.
[14]吴克寿,陈玉明,谢荣生,等.基于粗糙集与蚁群优化算法的特征选择方法研究[J].计算机应用研究,2011,28(7):2436-2438.Wu Keshou,Chen Yuming,Xie Rongsheng,et al.Rough sets and ant colony optimization based feature selection[J].Application Research of Computers,2011,28(7):2436-2438.
[15]杨鸿章.基于蚁群算法特征选择的语音情感识别[J].计算机仿真,2013(4):377-381.Yang Hongzhang.Feature selection of speech emotional recognition based on ant colony optimization algorithm[J].Computer Simulation,2013(4):377-381.
[16]尹宏鹏,刘兆栋,罗显科,等.一种基于粒子群优化的目标跟踪特征选择算法[J].计算机工程与应用,2013,49(17):164-168.Yin Hongpeng,Liu Zhaodong,Luo Xianke,et al.Target tracking feature selection algorithm based on particle swarm optimization[J].Computer Engineering and Applications,2013,49(17):164-168.
[17]姚旭,王晓丹,张玉玺,等.基于自适应t分布变异的粒子群特征选择方法[J].系统工程与电子技术,2013,35(6):1335-1341.Yao Xu,Wang Xiaodan,Zhang Yuxi,et al.Feature selection algorithm using PSO with adaptive mutation based t distribution[J].Journal of Systems Engineering and Electronics,2013,35(6):1335-1341.
[18]Storn R,Price K V.Minimizing the real functions of the ICEC 1996 contest by differential evolution[C]//Proceedings of IEEE International Conference Evolutionary Computer,1996:842-844.
[19]尹小娟,肖勤,琚报德.基于SOM的三维人脸表情聚类[J].信息系统工程,2013(3):137.
[20]孟海东,马娜娜,宋宇辰,等.基于密度函数加权的模糊C均值聚类算法研究[J].计算机工程与应用,2012,48(27):123-127.Meng Haidong,Ma Nana,Song Yuchen,et al.Research on fuzzy C-means clustering algorithm based on density function weighted[J].Computer Engineering and Applications,2012,48(27):123-127.
[21]郭均鹏,陈颖,李汶华.一般分布区间型符号数据的K均值聚类方法[J].管理科学学报,2013,16(3):21-28.Guo Junpeng,Chen Ying,Li Wenhua.K-means clustering of generally distributed interval symbolic data[J].Journal of Management Sciences in China,2013,16(3):21-28.
[22]潘锋,王建东,顾其威,等.基于图的特征选择算法[J].计算机工程,2012,38(9):197-198.Pan Feng,Wang Jiandong,Gu Qiwei,et al.Feature selection algorithm based on graph[J].Computer Engineering,2012,38(9):197-198.
[23]Liang J J,Suganthan P N,Deb K.Novel composition test functions for numerical global optimizationfC]//IEEE Swarm Intelligence Symposium 2005,Pasadena,California,2005:68-75.