基于LightGBM的特种纸打浆度预测模型研究与应用  

Research and Application of Special Paper Beating Degree Prediction Model Based on LightGBM

作　　者：刘迪成 马龙虎 周元洪 林海 王建 LIU Dicheng;MA Longhu;ZHOU Yuanhong;LIN Hai;WANG Jian(China Haisum Engineering Co.,Ltd.,Shanghai 200031,China;Winbon Schoeller New Materials Co.,Ltd.,Quzhou,Zhejiang 324000,China)

机构地区：[1]中国海诚工程科技股份有限公司,上海200031 [2]华邦古楼新材料有限公司,浙江衢州324000

出　　处：《自动化应用》2025年第7期278-281,284,共5页Automation Application

摘　　要：为解决特种纸行业打浆过程中纤维原料品种多、产品种类变化频繁造成的生产过程产品质量控制难题,基于LightGBM算法,建立打浆度预测模型。将模型采集过程变量和原料数据作为特征输入,输出打浆度预测值能满足实际生产的设定值控制要求,利用实际数据对模型进行了对比分析评价和验证。建立的数字孪生运行平台已在某特种纸公司成功应用,具有效果好、用时短、控制精度高的特点,有效降低了能耗,保障了产品质量。To solve the problem of product quality control in the special paper industry due to the variety of fiber raw materials and frequent changes in product types during the beating process,a beating degree prediction model is established based on the LightGBM algorithm.The process variables and raw material data collected by the model were used as feature inputs,and the predicted value of the beating degree was output to meet the set value control requirements of actual production.The model was compared,analyzed,evaluated,and validated using actual data.The established digital twin operation platform has been successfully applied in a special paper company,with the characteristics of good effect,short time,and high control accuracy,effectively reducing energy consumption and ensuring product quality.

关 键 词：LightGBM算法 打浆系统 数字孪生 

分 类 号：TS752[轻工技术与工程—制浆造纸工程]