摘要
为了提高铜转炉的操作水平,探讨了利用其生产运行中产生的大量数据建立优化决策模型的方法。针对过程数据含噪声、样本规模相对不足等问题,提出了一种鲁棒性更强的改进的神经网络建模方法;针对优化决策模型的应用目的,提出了支持度、置信度和相对置信度等模型评估指标;利用某厂的实际过程数据和前述方法,建立了基于神经网络的S1期(造渣1期)熔剂量和鼓风时间优化决策模型。实验效果表明所建优化决策模型能够显著改善S1期的吹炼效果。
To improve the operation level of copper converter,this paper studied the method of developing optimal decision-making model for copper-matte converting process.In view of the characteristics,such as containing noise,small sample size and so on,of the data produced in copper-matte converting process,proposed a new robust modelling method based on improved ANN(artifical neural network).Taking into account the application purpose of decision-making model,proposed some new indexes such as support degree,confidence degree,relative confidence degree to evaluate its validity.Using the historical data accumulated in production process of matte converting of a factory and methods mentioned above,developed optimal decision making models based on ANN for flux adding amout and blasting time of S1 period(that is the 1st slag making period).Simulation results show that these two decision-making models can significently improve the converting quality of S1 period.
出处
《计算机应用研究》
CSCD
北大核心
2011年第7期2539-2542,共4页
Application Research of Computers
基金
国家自然科学基金青年项目(60904077)
湖南省科技计划资助项目(2010FJ4132)
中国博士后科学基金资助项目(20100480950)
中央高校基本科研业务费专项资金(2011QNZT097)
关键词
铜锍转炉吹炼
数据挖掘
优化决策
神经网络
模型评估
copper-matte converting
data mining
optimal decision making
neural network
model evaluation