摘要
利用多元回归分析和人工神经网络构建螺旋分级机数学模型,预测校对分离粒度和分级精度,并对2个模型的预测效果进行了比较。使用LM算法和遗传算法优化人工神经网络模型,模型预测结果表明:采用人工神经网络构建螺旋分级机数学模型比多元回归分析模型表现出更高的拟合度和更低的拟合误差。
Mathematical models of spiral classifier were constructed based on multivariate regression analysis（MVRA）and artificial neural network（ANN）,and their predicted effects on the calibrated separation size and classification precision were compared. The ANN model was optimized by Levenberg-Marquardt（LM） algorithm and genetic algorithm.The results of model prediction testified that ANN model possesses higher fitting precision and less fitting error than MVRA model.
出处
《矿冶工程》
CSCD
北大核心
2017年第3期54-57,共4页
Mining and Metallurgical Engineering
基金
贵州省科学技术基金(黔科合JZ字[2014]2009号)
关键词
多元回归分析
人工神经网络
螺旋分级机
数学模型
multiple regression analysis
artificial neural network
spiral classifier
mathematical model
