神经网络与主元分析在采矿工程中的应用

Applization of combination of neural networks and principal component analysis in mining engineering

高质量的网络模型要求输入因素或输出因素之间尽可能不相关,以达到精度较高的预测模型的目的。建立了主成分分析法与神经网络结合的采矿方法优选模型,对神经网络的输入数据进行主成分分析。研究结果表明:该方法弥补了以往利用BP网络进行采矿方法优选过程中,由于输入数据相关使得输出数据精度下降的缺陷,使输入数据不相关,且减少了输入数据,消除了由于BP网络输入数据太多而影响数据处理速度的弊端,可使预测精度大大提高。

High-quality network model requests input or output factor irrelevant as far as possible in order to achieve the prediction model which has higher accuracy.An optimization model for mining was set up by using combination method of principal component analysis and neural networks,with the model.the principal components analysis of input-data of neural networks was make.The results show that through the method,the deficiencies were solved of lower output-data accuracy which was due to input-data relevant,during the process of using BP networks to optimize mining method in the past,thus the input-data was uncorrelated and reduced in input data quantity;and the defects were eliminated which was due to BP network input-data too much,influencd the data-processing speed;and the prediction accuracy of the model can be improved greatly.