Self-organizing feature map neural network and K-means algorithm as a data excavation tool for obtaining geological information from regional geochemical exploration data
CHEN Jun-Lin1, PENG Run-Min1, LI Shuai-Zhi1, 2, CHEN Xi-Cai2
1.School of Earth Sciences and Resources,China University of Geosciences (Beijing),Beijing 100083,China; 2.The First Geological Institute of The China Metallurgical Geology Bureau,Langfang 065201,China
Abstract：Regional geochemical data contain abundant geological information.The excavation of useful information from regional geochemical data is of important significance for the study of regional geology.In this paper,a model based on the self-organizing feature map and K-means algorithm is applied as a data excavation tool to discover hidden geological information from regional geochemical exploration data.For each data point,the raw data of each element is transformed by data normalization as the input value of the model.By means of SOM clustering and K-means clustering as the second step,the input data points can be divided into different groups,and then geological information can be acquired by analyzing the clustering results.Stream sediment survey data from southwest England is used as an example to test the performance of this model.The case study results demonstrate that:First,the clustering maps generated by the model agree well with the geological spatial distribution pattern.Accordingly,they can be used to predict the spatial distribution of geological bodies;Second,geological information is concealed in each cluster member.By analyzing and interpreting these geochemical characteristics, the geological information concealed in geochemical data can be discovered;Third,regional geochemical data excavation approach based on SOM network and K-means clustering is an effective geological information acquisition method,which can be used as a supplementary and strengthening way for conventional regional geology research.
[1] Rose A W,Hawkes H E,Webb J S.Geochemistry in mineral exploration[M].2 nd ed. London:Academic Press,1979. [2] Green P M.Digital image-processing of integrated geochemical and geological information[J].Journal of the Geological Society,1984,141(5):941-949. [3] Steenfelt A.Geochemical mapping and prospecting in Greenland —A review of results and experience[J].Journal of Geochemical Exploration,1987,29(1-3):183-205. [4] Shepherd A,Harvey P K,Leake R C.The geochemistry of residual soils as an aid to geological mapping:A statistical approach[J].Journal of Geochemical Exploration,1987,29(1-3):317-331. [5] Steenfelt A.Geochemical patterns related to major tectono-stratigraphic units in the Precambrian of northern Scandinavia and Greenland[J]. Journal of Geochemical Exploration,1990,39(1-2):35-48. [6] 刘德鹏,丁峰,汤正江.区域化探在森林沼泽区地质填图应用初探[J].物探与化探,2004,28(3):209-212,217. [7] 马晓阳,白显清,臧晓凡,等.黑龙江沙兰站幅森林沼泽区基础地质调查中的区域化探新方法[J].物探与化探,2005,29(2):108-110. [8] 史长义,任院生.区域化探资料研究基础地质问题[J].地质与勘探,2005,41(3):53-58. [9] 郝立波,陆继龙,马力.浅覆盖区土壤化学成分与基岩化学成分的关系及其意义——以大兴安岭北部地区为例[J].中国地质,2005,32(3):477 482. [10] 郝立波,陆继龙,李龙,等.区域化探数据在浅覆盖区地质填图中的应用方法研究[J].中国地质,2007,34(4):710-715. [11] 时艳香,郝立波,陆继龙,等.因子分类法在黑龙江塔河地区地质填图中的应用[J].吉林大学学报:地球科学版,2008,38(5):899-903. [12] Barnett C T,Williams P M.Using geochemistry and neural networks to map geology under glacial cover[R].Addendum #2 to Geoscience BC Report 2009-3.Boulder,Colorado:BWMINING,2010. [13] 向运川,龚庆杰,刘荣梅,等.区域地球化学推断地质体模型与应用——以花岗岩类侵入体为例[J].岩石学报,2014,30(9):2609-2618. [14] Sadeghi M,Billay A,Carranza E J M.Analysis and mapping of soil geochemical anomalies: Implications for bedrock mapping and gold exploration in Giyani area, South Africa[J].Journal of Geochemical Exploration,2015,154:180-193. [15] 高洪生,张全,曹淑萍,等.区域化探中利用三元图进行地质体划分及异常评价[J].物探与化探,2014,38(2):377-384. [16] 徐国志,徐锦鹏,段玲玲.化探资料在地质填图中的应用[J]. 物探与化探,2015,39(3):450-455. [17] Vriend S P,van Gaans P F M,Middelburg J,et al.The application of fuzzy c-means cluster analysis and non-linear mapping to geochemical datasets: examples from Portugal[J].Applied Geochemistry,1988,3(2):213-224. [18] Rantitsch G.Application of fuzzy clusters to quantify lithological background concentrations in stream-sediment geochemistry[J].Journal of Geochemical Exploration,2000,71(1):73-82. [19] Ghanbari Y,Hezarkhani A,Ataei M,et al.Regional geochemical pattern recognition with multivariate correspondence cluster analysis in the Ravar area, Iran[J].Applied Earth Science,2010,119(4):220-226. [20] Shiva M,Aryafar A,Zaremotlagh S.Fuzzy c-means cluster analysis, a robust multivariate technique in stream sediment geochemical exploration,a case study in Eastern part of Iran,Birjand[J].Journal of Geology and Mining Research,2011(1):1-6. [21] Abedi M,Norouzi G H,Torabi S A.Clustering of mineral prospectivity area as an unsupervised classification approach to explore copper deposit[J].Arabian Journal of Geosciences,2013,6(10):3601-3613. [22] Templ M,Filzmoser P,Reimann C.Cluster analysis applied to regional geochemical data: Problems and possibilities[J].Applied Geochemistry,2008,23(8):2198-2213. [23] Kohonen T.Self-organized formation of topologically correct feature maps[J].Biological Cybernetics,1982,43(1):59-69. [24] Kohonen T,Honkela T.Kohonennetwork[J].Scholarpedia,2007,2(1):83-100. [25] Wehrens R,Buydens L M C.Self-and Super-organizing Maps in R:The kohonenPackage[J].Journal of Statistical Software,2007,21(5):1-19. [26] Vesanto J,Alhoniemi E.Clustering of the self-organizing map[J].IEEE Transactions on Neural Networks,2000,11(3):586-600. [27] 栾龙源.基于自组织神经网络与模糊算法的彩色图像聚类分割系统[D].西安:西安电子科技大学,2010:1-56. [28] Sun Y.On quantization error of self-organizing map network[J].Neurocomputing,2000,34(1-4):169-193. [29] Kohonen T.Self-organizing maps[M].Berlin Heidelberg:Springer,2001. [30] Neme A,Miramontes P.Statistical properties of lattices affect topographic error in self-organizing maps[C]//Duch W,Oja E,Zadrozny S.Artificial Neural Networks:Biological Inspirations — ICANN 2005,15 th ,International Conference,Warsaw,Poland, September 11-15,2005,Proceedings,Part Ⅰ.Berlin Heidelberg:Springer,2005:427-432. [31] 周欢,李广明,张高煜.SOM+K-means两阶段聚类算法及其应用[J].现代电子技术,2010,33(16):113-116. [32] Hartigan J A,Wong M A.Algorithm AS 136:A k-means clustering algorithm[J].Journal of the Royal Statistical Society.Series C (Applied Statistics),1979,28(1):100-108. [33] Davies D L,Bouldin D W.A Cluster Separation Measure[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1979,PAMI-1(2):224-227. [34] Xie X L,Beni G.A validity measure for fuzzy clustering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1991,13(8):841-847. [35] Dunn J C.Well-Separated Clusters and Optimal Fuzzy Partitions[J].Journal of Cybernetics,1974,4(1):95-104. [36] 吴夙慧,成颖,郑彦宁,等.K-means算法研究综述[J].现代图书情报技术,2011,27(5):28-35. [37] Shail R K,Leveridge B E.The Rhenohercynian passive margin of SW England:Development,inversion and extensional reactivation[J].ComptesRendus Geoscience,2009,341(2-3):140-155. [38] Kirkwood C,Everett P,Ferreira A,et al.Stream sediment geochemistry as a tool for enhancing geological understanding:An overview of new data from south west England[J].Journal of Geochemical Exploration,2016,163:28-40. [39] 金秉福,林振宏,季福武.海洋沉积环境和物源的元素地球化学记录释读[J].海洋科学进展,2003,21(1):99-106. [40] 余烨,张昌民,李少华,等.元素地球化学在层序识别中的应用[J].煤炭学报,2014,39(s1):204-211. [41] 郑一丁,雷裕红,张立强,等.鄂尔多斯盆地东南部张家滩页岩元素地球化学、古沉积环境演化特征及油气地质意义[J].天然气地球科学,2015,26(7):1395-1404. [42] 刘英俊,曹励明,李兆鳞,等.元素地球化学[M].北京:科学出版社,1984:170-235. [43] Simons B,Shail R K,Andersen J C Ø.The petrogenesis of the early permian variscan granites of the cornubian batholith:lower plate post-collisional peraluminous magmatism in the rhenohercynian zone of SW england[J].Lithos,2016,260:76-94. [44] Beese A P.The argillite facies of the middle devonian succession in north cornwall[J].Proceedings of the Ussher Society,1982,5:321-332. [45] Tjhai G C,Furnell S M,Papadaki M,et al.A preliminary two-stage alarm correlation and filtering system using SOM neural network and K-means algorithm[J].Computers & Security,2010,29(6):712-723.
