论文部分内容阅读
针对矿产预测中已知矿点的样本数目较少的问题,该文提出了一种基于遗传算法优化的支持向量机矿产预测方法。采用遗传算法优化支持向量机的惩罚因子和径向基核函数参数,避免了参数选择不当对支持向量机预测结果的影响,从而提高矿产预测的精度。以空间建模工具ArcSDM中的卡林型金矿床数据为例进行实验。结果表明,支持向量机模型的预测准确率为89.3%,查准率为70.2%;而证据权方法的预测准确率为79.4%,查准率为50%,均小于支持向量机预测结果,说明遗传算法优化的支持向量机是一种有效的矿产预测方法。
Aiming at the problem of fewer sample numbers of known mine points in mineral prediction, this paper proposes a support vector machine (SVM) mineral prediction method based on genetic algorithm optimization. The genetic algorithm is used to optimize the penalty factor and RBF kernel parameters to avoid the influence of improper parameter selection on the prediction results of SVM, so as to improve the accuracy of mineral prediction. Taking the Carlin-type gold deposit data in ArcSDM, an example of space modeling tools, the experiment was carried out. The results show that the prediction accuracy rate of support vector machine model is 89.3% and the precision rate is 70.2%, while that of evidence-based method is 79.4% and the accuracy rate is 50% Genetic algorithm optimization support vector machine is an effective mineral prediction method.