Advertisement

Mine Water and the Environment

, Volume 38, Issue 4, pp 855–862 | Cite as

Combining the Fisher Feature Extraction and Support Vector Machine Methods to Identify the Water Inrush Source: A Case Study of the Wuhai Mining Area

  • Donglin Dong
  • Zhiyuan Chen
  • Gang LinEmail author
  • Xiang LiEmail author
  • Ruomeng Zhang
  • Yuan Ji
Technical Communication
  • 125 Downloads

Abstract

Discriminating the source of water inrush accurately and efficiently is necessary for water control in the coal mining industry. We combined the Fisher feature extraction and support vector machine (SVM) methods and applied this new model to the Wuhai mining area. The method extracts features from the raw data and integrated SVM, and synthetically considers the influence of geographical factors. Cross-analysis was tested 100 times, which arbitrarily selected 12 samples for the prediction and discrimination process. The results indicate that this new combined model of linear dimension reduction and non-linear dimension elevation was more accurate and efficient in discriminating water inrush sources than the traditional SVM model. Moreover, by reducing the penalty term of SVM model, we analyzed the correlation among the aquifers. We concluded that aquifers II and IV correlated strongly with each other, and that aquifer III was poorly connected with the other aquifers.

Keywords

Mine water-inrush source Intelligent recognition Coal mine 

Identifizierung der Wassereinbruchquelle mittels Kombination von Fisher-Merkmalextraktion und unterstützenden Vektormaschinen: Fallstudie im Wuhai-Bergbaugebiet

Zusammenfassung

In der Kohlenbergbauindustrie ist die präzise und effiziente Unterscheidung der Quelle von Wassereinbrüchen zur Kontrolle erforderlich. Wir kombinierten die Fisher-Merkmalextraktion (FFE) und die Methode unterstützender Vektormaschinen (SVM) und verwendeten das neue Modell im Wuhai-Bergbaugebiet. Die Methode extrahiert Merkmale aus Rohdaten und integrierten Vektormaschinen und bezieht den Einfluß geographischer Faktoren synthetisierend ein. Queranalysen wurden 100-fach getestet, daraus wurden 12 zufällig gewählte Proben für den Prozess der Vorhersage und Unterscheidung entnommen. Die Resultate zeigen, daß dieses neue kombinierte Modell der Reduktion linearer Dimensionen und nicht-linearer Höhendimension bei der Unterscheidung von Einbruchquellen genauer und effizienter war als das traditionelle SVM-Modell. Durch Reduktion des Strafterms des SVM-Modelles untersuchten wir die Korrelation der Aquifere. Wir schlossen daraus, daß die Aquifere I und IV stark miteinander korrelierten, während Aquifer III mit den anderen Aquiferen gering verbunden war.

Combinación de los métodos de extracción de características Fisher y de máquina de vectores de extracción para identificar la fuente de irrupción de agua: un estudio de caso del área minera de Wuhai

Resumen

Para el control de la irrupción de agua en la industria minera del carbón, es necesario discriminar la fuente de entrada de agua de forma precisa y eficiente. Hemos combinado los métodos de extracción de características Fisher (FFE) y de máquina de vectores de soporte (SVM) para aplicarlo posteriormente al área minera de Wuhai. El método extrae características de los datos en bruto y SVM integrado y considera sintéticamente la influencia de factores geográficos. El análisis cruzado se probó 100 veces, que seleccionó arbitrariamente 12 muestras para el proceso de predicción y discriminación. Los resultados indican que este nuevo modelo combinado de reducción de dimensión lineal y elevación de dimensión no lineal fue más preciso y eficiente para discriminar las fuentes de entrada de agua que el modelo SVM tradicional. Además, al reducir el término de penalización del modelo SVM, analizamos la correlación entre los acuíferos. Llegamos a la conclusión de que los acuíferos I y IV se correlacionaban fuertemente entre sí y que el acuífero III estaba pobremente conectado con los otros acuíferos.

Fisher特征提取与支持微量机(SVM) 相结合的矿井突水水源判别方法:以乌海矿为例

抽象

突水水源准确有效判别对矿井水害防治至关重要。构建了Fisher特征提取与支持向量机(SVM)相结合的突水水源判别方法,并运用于乌海煤矿。该方法从原始数据中提取特征要素并嵌入SVM模型,同时系统考虑了地理要素对水源判别的影响。交叉验证进行了100次,选出12组样本数据用以预测和识别。结果表明,与传统支持向量机(SVM)模型相比,新方法通过线性降维和非线性升维相结合使突水水源识别更准确、有效。另外,通过减少支持向量机(SVM)的惩罚项,分析了含水层之间联系。得出含水层I和含水层IV水力联系密,而含水层III与其它含水层联系较弱。

Notes

Acknowledgements

This study was financially supported by the Ministry of Science and Technology of China (2017YFC0804104), the Consultation Project of Chinese Academy of Engineering (2017-ZD-03-05-01), the National Natural Science Foundation (U1710258), and the China Postdoctoral Science Foundation (2018T110134).

Supplementary material

10230_2019_637_MOESM1_ESM.docx (24 kb)
Supplementary material 1 (DOCX 24 kb)

References

  1. Chen ZY, Zhang GZ, Wu CF, Yang SQ (2009) Application of water source identification in mine inflow based on support vector machines. Jiangxi U Sci Tech J 30:10–13 (In Chinese) Google Scholar
  2. Dong DL, Wu Q, Sun GM, Qian ZJ (2000) Research on water-filling patterns and countermeasures–a case study of Xiaotun coal mine. Int J Min Sci Tech 1:45–48Google Scholar
  3. Hu WW, Ma ZY, Cao HD, Liu F, Li T, Dou HP (2010) Application of isotope and hydrogeochemical methodsin distinguishing mine bursting water source. J Earth Sci Environ 32:268–271.  https://doi.org/10.3969/j.issn.1672-6561.2010.03.009 CrossRefGoogle Scholar
  4. Huang PH, Wang XY (2018) Piper-PCA-Fisher recognition model of water inrush source: a case study of the Jiaozuo mining area. Geofluids 5:1–10.  https://doi.org/10.1155/2018/9205025 CrossRefGoogle Scholar
  5. Huang YM, Zhang GB, Dong F, Da FP (2011) Multiple features extraction using Gabor wavelet transformation. Fisher faces and integrated SVM with application to facial expression recognition. Appl Res Comp 28:1536–1539 (In Chinese) Google Scholar
  6. Jiang SF (2015) Study on the prediction of the water in rush from coal seam floor during open coal mining-take a certain coal mine in Erdos as example. PhD diss, Inner Mongolia University of Science and Technology (In Chinese)Google Scholar
  7. Kwak N (2008) Feature extraction for classification problems and its application to face recognition. Pattern Recogn 41:1701–1717CrossRefGoogle Scholar
  8. Lee JJ, Uddin MZ, Kim TS (2008) Spatiotemporal human facial expression recognition using Fisher independent component analysis and hidden Markov model. Proc Int Conf IEEE Eng Med Biol Soc.  https://doi.org/10.1109/iembs.2008.4649719 CrossRefGoogle Scholar
  9. Ren H, Chang YL (2005) Feature extraction with modified Fisher’s linear discriminant analysis. In: Proceedings SPIE, vol 5995. Chemical and Biological Standoff Detection III, 599506 (4 November 2005), Boston, MA, United States.  https://doi.org/10.1117/12.631885
  10. Santhi P, Bhaskaran VM (2014) Detection of objects using fisher SVM with modified adaboost classification technique. J Theor Appl Info Tech 67:18–26Google Scholar
  11. Shan SG, Cao B, Gao W, Zhao DB (2002) Extended Fisherface for face recognition from a single example image per person. IEEE.  https://doi.org/10.1109/ISCAS.2002.1010929 CrossRefGoogle Scholar
  12. Tobler WR (1970) A computer movie simulating urban growth in the Detroit region. Econ Geogr 46(sup1):234–240.  https://doi.org/10.2307/143141 CrossRefGoogle Scholar
  13. Wang X, Ji HY, Wang Q, Liu XM, Huang D, Yao XP, Chen GS (2016) Divisions based on groundwater chemical characteristics and discrimination of water inrush sources in the Pingdingshan coalfield. Environ Earth Sci 75:872.  https://doi.org/10.1007/s12665-016-5616-3 CrossRefGoogle Scholar
  14. Wei LY (2013) Study on main problems and control countermeasures of safety production in Shenhua Wuhai coal mine. Shen Hua Sci Technol 11:3–5 (In Chinese)Google Scholar
  15. Yin NB, Hao J, Yu SC (2017) Research on establishment method of mine water source identificationmodel base based on artificial intelligence. J North Chin I Sci T 14:24–28 (in Chinese) Google Scholar
  16. Zhang H, Berg AC, Maire M, Malk J (2006) SVM-KNN: discriminative nearest neighbor classification for visual category recognition Proc. IEEE Conf on Comput Vis Pattern Recognit 2:2126–2136.  https://doi.org/10.1109/CVPR.2006.301 CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  1. 1.Department of Geological Engineering and EnvironmentChina University of Mining and Technology, Beijing (CUMTB)BeijingChina

Personalised recommendations