The Study On The Function Of Geological Variables In Mineral Resources Prediction

Nature of mineral resources prediction is a mathematical method to solve metallogenic prognosis. The solution to the problem is the change from geological problems into math problems, using mathematical methods to establish mathematical prediction model in order to predict the amount of resources. Modeling process is inseparable from geological data which extracts geological information that is to study geology carrier of geological variables. Therefore, through the process, geological variables occupy a key position in the geological variables. If there is closely relationship between geological variables studied and geological problems, adapting the mathematical methods, the research results are credible. When the variables are not related poorly to the study of geological problems, even if the mathematics is correct, nor are credible. Therefore, the geological variables can not be ignored in the digital geology major issue.Geological variables can be generally considered as the components and parameters of a mathematical model, and variables in the model performance play the model role in the overall assessment of capabilities which are called geological variable function. Geological variables in the interaction process can be subdivided into five classification that information conversion, level measurement, portfolio correlation, structural optimization. In my paper, I will discuss measure the extent of the main functions for variables, variable effective fields, the nature of the role and function of the direction of the new concept. The involved experimental data are under the "Eleventh Five-Year" National Technology Support Project "area Ailaoshan gold and polymetallic mineralization and exploration methods ". Now achievements in my paper are summarized as follows:1.General research of geological variables: First, collecting material on the basis of extensive literature; Second, summarizing the purpose of geological data preprocessing, meaning and method; Third, generally extract from the original geological data, Then select the principles and methods of geological variables; Lastly, systematically summarize geological variables concepts, classification and assignment. Accordingly, the paper proposes five functions of the geological variables: determine the classification of information conversion, level measurement, portfolio correlation, structural optimization.2.Orientation and role nature of geological variables: this paper gives the two concepts : orientation and role nature of geological variables. The author also provides the three direction of the geological variables and three types of properties, so as to find the role of the variable direction of the relationship between nature and role. In fact, the orientation of the variable is not isolated of a particular function of variable, but combined action based on the five functional meaning.3.Effective fields and information conversion of variable: through converting the information on the function of variables, The author proposes a concept -----variable effective fields. Variables, all possible values in its range, are playing a significant effect on the study that part of the interrelated functions of the range variable, which are called the effective domain. The signification of variable domain is the extraction of effective information and elimination interference information. Upon this basis, the paper uses a new method of discrete variables, namely, Monte Carlo France Moni, to seek effective domain method of quantitative discrete variables, ultimately gaining variables of information conversion. This paper also proposed using the minimum-dimensional space variable scale method of qualitative re-assignment, in order to achieve the variable information to high-level progressive, and then convert the information to achieve a variable function of purpose. 4.Research of function of the degree measure of variables: First, this paper introduces several conventional weighting methods for the geological variables. Eg: feature analysis,factor analysis and so on. In fact, to Measure the impact of variables on the target variable, we should consider both the importance of the variable itself and mutual constraints between variables (The feasibility of variables). Although variables affect the target as a group, we should excluding the comprehensive influence of the variables. The importance of the variable is impact of a variable on the target by itself. To know the importance of the variable, we can measure the status of variables. And we have to consider the feasibility of the variable to know if the importance of the variable can be achieved in the interaction process, because Interactions between variables constraints the importance of the variable. Therefore, a new method used for screening geological variables is presented in this paper. By Conventional method to measure the importance of variables, each variable is assigned with a fixed " weight ", in the new method named Two-dimensional weighting method, accuracy(precision) of geological variables in the resource evaluation model will be improved by decomposing the variable weight into important weight and feasible weight. With the Two-dimensional weighting method, we can establish More stable statistical model to predict the predicated geological problem better.Variable's important weight is a scale of relatively importance for every variable on the condition that variable aggregation wound affect the target, that is to say, it is a scale of variable'prediction ability to targets exclude other variable. Variable'feasible weight is the link/relation among the variable, it is a scale of effect among variables. The author give out the result of important weight and feasible weight about the series variable and two states variable in the paper, and define variable's coefficient of partial correlation et c to denote the important weight of series variable, variable's correlation coefficient to denote the feasible weight of variable. We also define coefficient of partial correlation,optimization prediction coefficient, contingency coefficient, information quantity et c to denote the important weight of two states variable, and associate coefficient, similarity coefficient, match coefficient et c to denote the feasible weight. Because of the differences in degree and combination, bound relationship between variable's important weight and feasible weight, we combine both weights to be a synthesis weight by mathematic model and consider their impact factors first, and compute weight coefficient by a mathematic programming model, then we obtain the synthesis weight. The paper gives out three mathematic models for solving the synthesis weight.5.In this example, the author calculates the binary-state variable as the possible weight of the variable, takes the contingency coefficient between variable and dependent variable as important weight, fixes the impact factors of weight(α,β) by mathematical model, then further defines the variable's synthesis weight coefficient. Comparing variable and its corresponding metallogenetic connection degree of each unit selected by approach of synthetic weight with variable and metallogenetic unit selected by haracteristic analysis approach, the result shows that it is practical and feasible that establishing prediction model by screening geological variable via assigning synthetic weight for geological variables. In some cases, the assignment of a variable using single weight method may result in the lost of information in computing unit relation degree. Whereas using the synthesis weight is a more reasonable way to select the variable because the synthesis weight demonstrate the importance of variable to target and relationships among variable, so the computation result of unit relation degree through synthesis weight can reflect the importance and connection degree of the variable, and the computation result is more perfect and reliable.