{"title":"Multiobjective Optimization of Antenna Inverse Design With Data Augmentation Based on K-Means-NN","authors":"Meng Wang;Shirui Yu;Jian Dong;Heng Luo;Chengwang Xiao","doi":"10.1109/LAWP.2025.3584825","DOIUrl":null,"url":null,"abstract":"To enhance the accuracy of the inverse model with a limited number of antenna sample points, a data augmentation method based on K-means and forward neural networks (FNNs) is proposed. In this method, K-means is used to partition the initial imbalanced dataset, while a perturbation factor and FNNs are introduced to oversample the minority samples, ensuring a balanced performance distribution. A weighted loss combining soft dynamic time warping and mean squared error is embedded in FNNs to predict antenna performance more accurately. The convolutional inverse neural network integrates multiple performance features stacked along a specific dimension and is iteratively invoked until the optimized antenna meets the desired objectives. Taking a wideband circularly polarized antenna as a numerical example, the proposed method requires fewer electromagnetic simulations than other advanced optimization techniques.","PeriodicalId":51059,"journal":{"name":"IEEE Antennas and Wireless Propagation Letters","volume":"24 9","pages":"3144-3148"},"PeriodicalIF":4.8000,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Antennas and Wireless Propagation Letters","FirstCategoryId":"94","ListUrlMain":"https://ieeexplore.ieee.org/document/11061809/","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0
Abstract
To enhance the accuracy of the inverse model with a limited number of antenna sample points, a data augmentation method based on K-means and forward neural networks (FNNs) is proposed. In this method, K-means is used to partition the initial imbalanced dataset, while a perturbation factor and FNNs are introduced to oversample the minority samples, ensuring a balanced performance distribution. A weighted loss combining soft dynamic time warping and mean squared error is embedded in FNNs to predict antenna performance more accurately. The convolutional inverse neural network integrates multiple performance features stacked along a specific dimension and is iteratively invoked until the optimized antenna meets the desired objectives. Taking a wideband circularly polarized antenna as a numerical example, the proposed method requires fewer electromagnetic simulations than other advanced optimization techniques.
期刊介绍:
IEEE Antennas and Wireless Propagation Letters (AWP Letters) is devoted to the rapid electronic publication of short manuscripts in the technical areas of Antennas and Wireless Propagation. These are areas of competence for the IEEE Antennas and Propagation Society (AP-S). AWPL aims to be one of the "fastest" journals among IEEE publications. This means that for papers that are eventually accepted, it is intended that an author may expect his or her paper to appear in IEEE Xplore, on average, around two months after submission.