摘要
该文在进行某选矿厂浮选生产数据分析的过程中,针对浮选过程常常为高度非线性多输入多输出问题的特点,在深入剖析BP网络与RS理论的基本原理和特点的基础上,提出了分别利用BP网络的高度非线性拟合特性对浮选生产数据进行训练以获得浮选生产过程知识的隐式表达,同时利用RS理论的数据浓缩功能对浮选生产数据进行约简而得到相应浮选生产过程知识的显式表达,然后对两种模型的分析结果进行交叉验证的应用模式。与基于人工神经网络的决策树构造等其它人工神经网络的白化方法相比,该方法具有在保证问题分析结果的精度的同时,分析过程相对简单,克服了由于BP网络结构的不确定性而导致最终得到的决策树不确定的缺点,并由此减小了对所分析数据产生误解的风险。
With the analysis of the flotation data of a mineral processing plant ,after the brief review of the basic principles and characteristics of BP net and rough set theory,a new method is put forward,which is suitable for the nonlinear regression of the complex flotation data to get the explicit rules by making use of rough set theory to enrich the flotation data to get the law of flotation as a complementary tool of BP net,and then the crossed test can be made.This method is simpler than other methods such as establishing decision tree based on neural network and overcomes its limitation of the uncertainty of the established decision tree due to the uncertainty of the BP net architecture and reduces the risk of misunderstanding the data to analyze.
出处
《计算机工程与应用》
CSCD
北大核心
2002年第7期218-220,共3页
Computer Engineering and Applications
基金
河北省教委博士科研启动基金资助