The application of GRNN and LS-SVM to coal properties calculation
ZHOU Da-Peng1, WANG Zhu-Wen1, LI Xiao-Chun2
1. College of Geoexploration Science and Technology, Jilin University, Changchun 130026, China;
2. Coal Geological Bureau of Inner Mongolia, Hohhot 010000, China
As one of the most pivotal resources, coal cannot be replaced. The evaluation of coal properties plays an essential role in the development. The calculation of coal properties based on laboratory analysis is inefficient and expensive. In this paper, the authors have resolved this problem by establishing the relationship between logging parameters and coal properties. Natural gamma, time difference, density and three-lateral resistivity are treated as input, and values of moisture, ash, volatile matter and fixed carbon are chosen as output. By using 73-layer logging data to train, the authors constructed a model based on GRNN and LS-SVM to calculate coal qualities. Through testing 19-layer data, the authors have reached the conclusion that these two methods can be well used in practice. The GRNN can calculate the content of moisture, ash, volatile matter and fixed carbon more accurately than LS-SVM, with its mean square error lower than 1%.
周大鹏, 王祝文, 李晓春. GRNN与LS-SVM方法在计算煤质工业组分中的应用[J]. 物探与化探, 2016, 40(1): 88-92.
ZHOU Da-Peng, WANG Zhu-Wen, LI Xiao-Chun. The application of GRNN and LS-SVM to coal properties calculation. Geophysical and Geochemical Exploration, 2016, 40(1): 88-92.