论文部分内容阅读
在深入研究花岗岩型铀矿成矿模式的基础上,本文首次将模式识别中的Cora-3方法应用于成矿预测。结果表明:从对象的确定、特征的提取和选择,到学习、投票等一整套Cora-3模式识别算法对铀矿远景预测是可行的,有效的。删除特征试验可得出控制花岗岩型铀矿的主要地质特征。控制试验的投票结果所得出的成矿远景与地质勘探、地质预测是一致的。
Based on the deep research on the metallogenic model of granite-type uranium deposits, the paper first applies the Cora-3 method in pattern recognition to metallogenic prediction. The results show that a complete set of Cora-3 pattern recognition algorithms are feasible and effective for the prediction of uranium prospect from the object identification, feature extraction and selection, to learning and voting. Deletion of characteristic tests leads to the control of the main geological features of granitic uranium deposits. The results of the control experiment show that the prospect of mineralization is consistent with the geological prospecting and geological prediction.