模型预测控制技术在婴配乳粉配料优化中应用  

Application of Model Predictive Control Technology in Optimizing of the Infant Milk Powder Ingredients

作　　者：雷霆 孙忱 王潘文 王博 童邦彦 刘少莉 储小军 何光华 LEI Ting;SUN Chen;WANG Panwen;WANG Bo;TONG Bangyan;LIU Shaoli;CHU Xiaojun;HE Guanghua(School of Biological and Chemical Engineering,Zhejiang University of Science and Technology,Hangzhou,Zhejiang 310023;Zhejiang Yiyuansu Nutrition Technology Co.,Ltd.,Quzhou,Zhejiang 324000;Beingmate(Hangzhou)Food Research Institute Co.,Ltd.,Hangzhou,Zhejiang 310053)

机构地区：[1]浙江科技学院生物与化学工程学院,浙江杭州310023 [2]浙江益元素营养科技有限公司,浙江衢州324000 [3]贝因美(杭州)食品研究院有限公司,浙江杭州310053

出　　处：《核农学报》2024年第8期1532-1538,共7页Journal of Nuclear Agricultural Sciences

基　　金：浙江省教育厅科研项目资助(Y202248510);科研成果奖培育重点团队项目(2023JLZD009);杭州市重大科技专项(202203A11)。

摘　　要：为降低婴幼儿配方乳粉生产过程中的营养成分含量波动,本研究基于模型预测控制(MPC)算法和序列二次规划(SQP)算法构建婴幼儿配方乳粉配料优化控制模型,进行配料过程营养成分含量预测、反馈校正和滚动优化。结果表明,通过配料优化控制模型模拟干预100批次乳粉生产过程后,蛋白质、脂肪和碳水化合物含量的标准差分别从5.18、4.91、5.86 g·kg^(-1)降低至1.01、1.22和1.33 g·kg^(-1),验证了配料优化控制模型的有效性。本研究可为实现婴幼儿配方乳粉的配料优化控制提供参考。In order to reduce the fluctuation of nutrient content during the production of infant formula milk powder,a formulation optimization control model based on the model predictive control(MPC)algorithm and sequential quadratic programming(SQP)algorithm was proposed to realize the prediction,feedback correction and rolling optimization of nutrient content during the formulation process.Our approach was employing employing the formulation optimization control model to 100 batches of milk powder production processes.Results showed that the standard deviations of protein,fat and carbohydrate content decreased from 5.18,4.91 and 5.86 g·kg^(-1) to 1.01,1.22 and 1.33 g·kg^(-1),respectively.Thus,the effectiveness of the established formulation optimization control model was verified.This study could serve as a reference for achieving formulation optimization control in the production of infant formula milk powder.

关 键 词：婴幼儿配方乳粉 配料优化控制 模型预测 序列二次规划 

分 类 号：O231[理学—运筹学与控制论] TS252.51[理学—数学]