圆柱壳声学超材料带隙的机器学习优化设计  

Machine Learning Optimization Design of Acoustic Metamaterial Band Gaps for Cylindrical Shells

作　　者：陈思危 姚敦辉 姚凌云[1] CHEN Siwei;YAO Dunhuei;YAO Lingyun(College of Engineering and Technology,Southwest University,Chongqing 400715,China)

机构地区：[1]西南大学工程技术学院,重庆400715

出　　处：《噪声与振动控制》2024年第4期70-76,共7页Noise and Vibration Control

基　　金：国家自然科学基金资助项目(52175121)。

摘　　要：圆柱壳结构广泛应用于工程中,其振动噪声水平对工程设计有着巨大的影响。针对圆柱壳声学超材料的带隙设计要求,提出一种基于机器学习理论的带隙优化方法。首先,利用有限元法对3000个圆柱壳超材料胞元结构进行带隙分析,然后将计算的带隙数据分别输入不同的机器学习模型进行训练,构造预测模型并利用决定系数(R2),均方误差(Mean Square Error,MSE),可解释方差(Explained Variance Score,EVS)等指标寻求其最优模型,最后,利用最优机器学习模型对圆柱壳声学超材料的带隙进行拓宽设计。研究结果显示该方法可对圆柱壳声学超材料带隙进行有效预测与拓宽。Cylindrical shell structures are widely used in engineering,and their vibration and noise levels have a great impact on engineering design.According to the requirements of band gap design of cylindrical shell acoustic metamaterials,a method of band gap optimization based on machine learning theory is proposed.Firstly,the finite element method is used to analyze the band gaps of 3000 cylindrical shell metamaterial cell structures,and the calculated band gap results are input into different machine learning models for training.Then,the prediction model is constructed and the best model is searched according to the indicators such as the coefficient of determination(R2),Mean Square Error(MSE)and Explained Variance Score(EVS).Finally,the optimal machine learning model is used to broaden the band gap of cylindrical shell acoustic metamaterials.The research results show that this method can effectively predict and broaden the band gap of cylindrical shell acoustic metamaterials.

关 键 词：声学 圆柱壳 声学超材料 机器学习 带隙 优化 

分 类 号：X593[环境科学与工程—环境工程]