基于WOE⁃GA⁃BP神经网络模型对陆相火山岩型铜矿成矿预测研究——以宁芜盆地(江苏部分)为例

Metallogenic Prediction of Continental Volcanic Copper Deposits Based on WOE⁃GA⁃BP Neural Network Model:Taking Ningwu Basin(Jiangsu Part)as an Example

宁芜盆地位于长江中下游铜、金、铁、铅、锌、硫、石膏成矿带,构造运动强烈、岩浆活动频繁,成矿地质条件优越.通过提取宁芜盆地(江苏部分)地层、断裂构造、航磁、化探异常等9个控矿因子基础信息,采用证据权重与基于遗传优化的BP神经网络(WOE⁃GA⁃BP)模型,对该区域内陆相火山岩型铜矿进行成矿预测研究,使用混淆矩阵及ROC曲线进行模型精度评价.经叠加分析,基于成矿模型圈定的铜矿A、B、C类远景区与江苏省重要矿产潜力评价圈定的铜矿A、B、C类预测区分别有81.23%、62.69%、100%的重叠区,表明本次预测结果较为可靠,为区域成矿预测提供了新的思路和方法,对后续勘查工作具有一定的指导意义.

Ningwu Basin is located in the copper,gold,iron,lead,zinc,sulfur,and gypsum mineralization belt in the middle and lower reaches of the Yangtze River,with strong tectonic movement,frequent magmatic activity,and superior geological conditions for mineralization.By extracting the basic information of 9 ore⁃controlling factors such as strata,fault structures,aeromagnetic,and geochemical anomalies in Ningwu Basin(Jiangsu part),a coupling model of weight of evidence and BP neural network model based on genetic optimization(WOE⁃GA⁃BP)is used to conduct mineralization prediction research on continental volcanic rock copper deposit in the region,the confusion matrix and ROC curve are used to evaluate the model accuracy.After overlay analysis,there are 81.23%,62.69%,and 100%overlap areas between the copper deposit A,B,and C prospect areas delineated based on the mineralization model and the copper deposit A,B,and C prediction areas delineated by the evaluation of important mineral potential in Jiangsu province,respectively.The results getting by this research indicate that the prediction results are relatively reliable and provide new ideas and methods for regional mineralization prediction,which has certain guiding significance for subsequent exploration work.