ID3决策树在预测电解槽出铝量中的研究与实现

Research and realization of ID3 decision tree in predicting aluminum output of aluminum pots

人工控制决策方式已经难以适应现代铝电解生产要求,知识自动化是实现铝电解槽寻优决策的必由之路,铝电解生产系列中数百台电解槽会产生海量的结构化生产数据,蕴含有丰富的知识。通过与人工决策的非结构化经验知识进行整合,可实现基于生产大数据的知识自动获取和数据-知识集成控制。文中提出一种改进的ID3算法,应用回归分析计算各条件属性影响出铝量的权重,对铝电解数据库中包含的出铝量专家知识和经验进行知识表示和自动推理,算法采用Python语言实现,从而生成出铝量决策规则,辅助工艺管理人员做出科学判断,提高生产智能管理水平。

The artificial control decision-making method has been difficult to adapt to the requirements of modern aluminum electrolysis production,and the knowledge automation is the only way to realize the optimization decision-making of aluminum pots.Hundreds of aluminum pots in the aluminum potline will generate massive structured production data,which contains rich knowledge.By integrating with the unstructured experience knowledge of artificial decision-making,the automatic knowledge acquisition and data knowledge integrated control based on production big data can be achieved.In this paper,an improved ID3 algorithm is proposed.Regression analysis is used to calculate the weight of each condition attribute affecting the aluminum output.Knowledge representation and automatic reasoning are carried out for the aluminum output expert knowledge and experience contained in the aluminum electrolysis database.The algorithm is implemented in Python language,so as to generate aluminum output decision rules and assist process managers to make scientific judgments,as well as improve the level of intelligent production management.