摘要
采用多源信息找矿模型,结合Kohonen人工神经网络对新疆东天山地区斑岩型钼矿进行了成矿预测。通过该地区已有斑岩型钼矿的成矿、控矿规律,确定了5类预测变量。由于东天山钼矿床已知样本较少,使用非线性的Kohonen人工神经网络法进行少模型预测。此方法不依赖预测区域的样本数量,实行非监督分类。分类结果显示:东天山地区2个典型钼矿床皆落入A类成矿有利区域,证明分类效果较为可信。实验结果表明,该方法操作简便,是一种较为快捷、有效的预测方法。
Based on Kohonen artificial neural networks, muhi-information ore prospecting model is applied in the prediction of porphyry molytbdenmn deposits in East Tian Shan. Through the ore-fnrming and ore-controlling regularities of the molybdenum deposits, 5 pre- dietion variables were determined. For there are few known samples of molybdenum deposits in East Tian Shau, we used the nonlinear Kohonen artificial neural uetworks method. This method does not rely on the numbers of predietion samples with the unsupervised clas- sification, aml the elassifieation result shows that two typical molybdenum deposits both fall in the A categoh＇y metallogenetie prospective area, proving that this classification is more eredible. The experimental results show that the method is simple, convenient and more efficient.
出处
《地质学刊》
CAS
2015年第2期231-235,共5页
Journal of Geology
基金
中国地质调查局项目"全国重要成矿区带矿产区划部署综合研究"(12120114051401)
"重点成矿区带矿产资源综合评价与区划"(12120113092700)联合资助
关键词
Kohonen人工神经网络
钼矿
预测
新疆东天山
Kohonen artificial neural network
molybdenum deposits
metallogenic prediction
East Tian Shan, Xinjiang
