Mining and Analysis of Mineral Information in Gold Mine Based on Rough Set Theory

Abstract: At present, the mineral data are characterized by time-space and multi sources and the means for acquiring mineral information are diversiform, which make the mineral resources exploration, and evaluation increasingly complicated. GIS technology has provided the capability of integrated management, query and research of multi-source geological information. Computer science and mathematical geology is the technical support for optimal selection of target and comprehensive analysis. The exploration of mineral information and quantitative analysis of multivariate geological information are the trend in future development.Geological phenomena and ore-forming process have inner complexity, which has transcended the confines of linear means. The sampling observation in deposit research is random, the information extract is deficient, and the linear prognosis model is limit, which make the research result in geology impossible to be accurately recurrence, so that most deduced prognosis have multi solutions. Therefore, it's necessary to study on the application of nonlinear theory and the extract means of faint information, which is a hotspot in qualitative and quantitative geological research.In mineral exploration research, the field data collecting and the indoor information disposing are affected by human thinking mode. Rough Set can combine with qualitative and quantitative genes, independent of any prior knowledge, absolutely aim at data to evaluate the contribution ratio of mineral genes scientifically, extract the relation and rules between geological genes through numerous data, and provide scientific basis for prospecting evidences.The work is supported by National Science Foundation of China. Based on geology and metallogenic regulation, make use of mathematic tools and GIS technique, and analyze geological variables and metallogenic probability quantitatively based on Rough Set to get best combination of variables. Apply characteristic analysis and Neural Network to establish logical prospecting model for complex geological problem with multi goals and multi genes, analyze deposit variable quantitatively, to meet the goal of target optimization.Multi-source data are integrated on GIS platform. Considering metallogenic background, data disposal environment and data quality, essential mineral information of geological cells are extracted to establish decision table. Attributes are reduced through Rough Set to get the attribute core. Based on the reduced attribute table, approximate set is acquired through Variable Precision Rough Set model. Certain rules are acquired from lower approximation and the attribute values are reduced by applying decision matrix. The reduced result show good agreement with practice.The main idea of Rough Set is the expression and reduction of knowledge and the main function is to find out the mineral prospecting information of geological evidence. The combination of Rough Set and characteristic analysis indicates the error of the model establish by reduced variables is minor. The combination of Rough Set and Neural Network can reduce complexity of the network and simplify the model. Results of two models are in substantial agreement, which indicates the integrated prognosis method based on Rough Set has a certain reference value for metallogenic prognosis. The method is worth further research to inherit and develop the mathematic geology…
Key words: Rough Set; mineral information; quantitative prognosis; characteristic analysis; neural network

This entry was posted in Master Thesis. Bookmark the permalink.