基于回归的软化水系统运行时间预测模型研究  

Study on Prediction Model of Run Time of Water Softening System Based on Regression

作　　者：代军 李昕 马瑞 王献 杨春明[2] 

机构地区：[1]四川中烟工业有限责任公司绵阳卷烟厂,四川绵阳621000 [2]西南科技大学计算机科学与技术学院,四川绵阳621010

出　　处：《工业控制计算机》2022年第6期11-13,16,共4页Industrial Control Computer

摘　　要：基于钠离子交换器的软化水系统运行时间主要根据经验人工设置,在来水中钙镁离子含量变化较大时,很难保证生产的软化水质量。分析了软化水系统中各生产环节电导率数据的相关性,采用了岭回归、CART回归和多项式回归三种机器学习预测模型,对软化水的生产时间进行了建模分析,并通过预测误差、运行时间、误差下的预测准确率对模型进行了评估。实验的结果表明,Ridge回归的误差最小,运行时间少,误差允许范围内的准确率最高。The run time of the water softening system based on sodium ion exchanger is mainly set manually according to experience.When the quantity of calcium and magnesium ions in the incoming water changes greatly,the quality of the produced softened water cannot meet the standard.This paper analyzes the correlation of the conductivity data of each pro-duction process in the softened water system,three machine learning prediction models,Ridge,CART and Polynomial,and ana-lyzes the run time of water softening. The model is evaluated by prediction error,running time and prediction accuracy under error.The experimental results show that ridge regression has the smallest error,less running time and the highest accuracy within the allowable error range.

关 键 词：钠离子交换 回归模型 时间预测 机器学习 

分 类 号：TU991.2[建筑科学—市政工程]