The Research And Realization Of Coal Quality Prediction System Based On Geostatistics

The prediction of coal resources based on distribution, reserves and grade becomes a hotspot of coal geological research work. The exploration of their spatial distribution and the improvement of prediction accuracy is very important. Spatial interpolation is the basic and key method of the Coal Quality Prediction, and the scientificity and rationality of the interpolation model directly affect the level of prediction accuracy. However, there are larger estimation error and even systematic bias in the traditional interpolation methods; and the degree of automation of the model parameters fitting in the geostatistical methods is not high, which leads to the decline of simplicity and popularity of the commercial software. Therefore, to achieve the automation of fitting the model parameters and to display the spatial distribution and trend of coal quality by graphics must have important practical significance and application value.Firstly, this paper introduces the research contents of Geostatistical and their superiority over the traditional estimation methods and Classical Statistics methods in the common questions about estimation of resources reserves and summarizes the characteristics and applicable conditions of Geostatistics methods. And then, it elaborates Regionalized Variables, Semi-Variogram, Estimation Variance and other basic theory of Geostatistics, and in-depth analysis of the Ordinary Kriging valuation model and its superiority--unbiasedness and optimality.Secondly, comparing the Semi-Variogram and the calculating methods of Relative Semi-Variogram, this paper also summarizes the factors that affect the accuracy of Variogram Functions(such as the choice of regionalized variables, effective treatment of specific values, determination of the amount of data, etc) and compares the two applicable results of the Weighted Linear Programming Method and regular Variogram Fitting Method. After that, to achieve the automation of spherical model parameter fitting, Kriging Spherical Model is fitted by the Weighted Linear Programming Method, the optimization of which is also tested by Cross-Validation Method and Integrated Indicator Method.Thirdly, the paper introduces the drawing principles of Java 3D and the development process of the programs. By Servlet, Embeding Java 3D Applet into webpages implements access to the database. And using the digital signature and cut-off screen technology implements the preservation of Java 3D graphics, which leads to the realization of printing the geological maps.Finally, using Oracle 9i Database, this system establishs the mine geology and coal quality database with J2EE programming framework. Meanwhile, using the structure of Java 3D scene graph, the modeling of the geological map also describes the drawing process of the rule-based grid contour. The design and realization of the Coal Quality Prediction system validates the feasibility and effectiveness of the research results in this paper.