残差分析在河南嵩县庙岭金矿模型验证中的应用研究

Application of residual analysis in model validation for the Miaoling gold deposit in Song County,Henan Province

残差分析是通过分析已知值与预测值之差来检验模型准确性的统计学方法。本文以河南省嵩县庙岭金矿为例,对矿体块模型进行残差分析,用以验证模型的准确性和可靠性。首先,收集矿区勘查工程分析数据,构建钻孔三维模型、矿体实体模型和块模型。其次,利用剖面观察验证、整体分布验证和切片验证等方法对矿体块模型进行初步验证。最后,利用残差分析的方法对矿体块模型进一步验证。勘查工程数据和块模型的残差分析显示,残差平均值为-0.049×10^(-6),直方图总体呈正态分布,平均值处于累计分布50%附近,说明矿体块模型精确性和可靠性较高。同时,残差均值为负数,偏态系数为0.231,直方图负值位置偏高,累计分布中负值25%处和对应的正值没有对称分布,宽度大于正值,显示模型估值仍存在不足。为进一步优化模型,依据勘查工程样品空间的分布特征,设置6套不同的估值参数进行实验,并分别对比残差平均值、偏态系数和直方图形态,使建模估值参数达到最优。

Residual analysis is a statistical technique utilized to assess the accuracy of a model by examining the disparity between the known value and the predicted value.Using the Miaoling gold deposit in Song County,Henan Province as a case study,this paper conducts a residual analysis on the ore block model to validate its accuracy and reliability.Firstly,the exploration engineering analysis data in the mining area were collected to establish a three-dimensional drillhole model,an ore body solid model,and a block model.Secondly,initial validation of the ore body block model was conducted through methods such as profile observation validation,overall distribution validation,and slice validation.Finally,the residual analysis method was employed to further verify the accuracy of the ore body block model.The residual analysis of the exploration engineering data and the block model revealed an average residual of-0.049×10^(-6),while the histogram exhibited a predominantly normal distribution,with the mean value closely aligned to the 50%cumulative distribution.It indicates a high level of accuracy and reliability of the ore body block model.Meanwhile,the mean of the residuals is negative,with a skewness coefficient of 0.231.The histogram shows a higher proportion of negative values,and the cumulative distribution does not exhibit a symmetrical distribution between the 25%percentile of negative values and their corresponding positive values.The width of the distribution for negative values is greater than that for positive values,indicating that there are still deficiencies in the model's estimation.To further optimize the model,based on the spatial distribution characteristics of the exploration engineering samples,six different sets of estimation parameters have been set for experimentation.The average residual,skewness coefficient,and histogram patterns have been compared respectively to achieve optimal modeling estimation parameters.