轻工机械2026，Vol.44Issue(2)：1-10,10.DOI:10.3969/j.issn.1005-2895.2026.02.001
基于深度学习与遗传算法的铝罐二重卷封工艺参数多目标优化
Multi-Objective Optimization of Aluminum Can Double Seaming Process Parameters Based on Deep Learning and Genetic Algorithm
摘要
Abstract
The double seaming process of aluminum cans is a critical manufacturing procedure that determines the sealing reliability and operational safety of beverage packaging.Its forming quality directly affects the product's leakage resistance and the service life of production equipment.To synergistically improve the overlap rate and tightness while effectively controlling roller contact stress,a multi-objective optimization method integrating the Deep Feedforward Neural Network(DFNN)and Non-dominated Sorting Genetic Algorithm Ⅱ(NSGA-Ⅱ)was proposed.First,a high-precision finite element model was established to analyze the influence mechanisms of key process parameters on sealing performance and mechanical response.Then,a DFNN surrogate model was constructed to achieve accurate mapping of the overlap rate,wave peak count and maximum contact stress.Finally,the NSGA-Ⅱ algorithm was adopted for multi-objective optimization to obtain the Pareto optimal process parameter combination.The research results show that the optimized scheme is significantly superior to the initial design in terms of overlap rate,tightness and maximum contact stress,and the error between the experimental verification results and the simulation prediction results is less than 7％.The proposed method is effective and engineering applicable for comprehensively improving sealing performance and extending the service life of equipment.关键词
饮料包装/铝罐/二重卷封/多目标优化/深度学习/遗传算法Key words
beverage packaging/aluminum can/double seaming/multi-objective optimization/deep learning/genetic algorithm分类
轻工纺织引用本文复制引用
张子璇,杨建余,于培师,赵军华..基于深度学习与遗传算法的铝罐二重卷封工艺参数多目标优化[J].轻工机械,2026,44(2):1-10,10.基金项目
国家自然科学基金(12372078) （12372078）
江苏省基础研究重点项目(BK20243044). （BK20243044）
