|
|
Frontiers of Information Technology & Electronic Engineering
ISSN 2095-9184 (print), ISSN 2095-9230 (online)
2023 Vol.24 No.6 P.916-926
Ka-band broadband filtering packaging antenna based on through-glass vias (TGVs)
Abstract: This work presents a novel design of Ka-band (33 GHz) filtering packaging antenna (FPA) that features broadband and great filtering response, and is based on glass packaging material and through-glass via (TGV) technologies. Compared to traditional packaging materials (printed circuit board, low temperature co-fired ceramic, Si, etc.), TGVs are more suitable for miniaturization (millimeter-wave three-dimensional (3D) packaging devices) and have superior microwave performance. Glass substrate can realize 3D high-density interconnection through bonding technology, while the coefficient of thermal expansion (CTE) matches that of silicon. Furthermore, the stacking of glass substrate enables high-density interconnections and is compatible with micro-electro-mechanical system technology. The proposed antenna radiation patch is composed of a patch antenna and a bandpass filter (BPF) whose reflection coefficients are almost complementary. The BPF unit has three pairs of λg/4 slots (defect microstrip structure, DMS) and two λg/2 U-shaped slots (defect ground structure, DGS). The proposed antenna achieves large bandwidth and high radiation efficiency, which may be related to the stacking of glass substrate and TGV feed. In addition, the introduction of four radiation nulls can effectively improve the suppression level in the stopband. To demonstrate the performance of the proposed design, a 33-GHz broadband filtering antenna is optimized, debugged, and measured. The antenna could achieve |S11|<-10 dB in 29.4‒36.4 GHz, and yield an impedance matching bandwidth up to 21.2%, with the stopband suppression level at higher than 16.5 dB. The measurement results of the proposed antenna are a realized gain of ~6.5 dBi and radiation efficiency of ~89%.
Key words: Filtering packaging antenna (FPA); Through-glass vias (TGVs); 3D packaging devices; Laser bonding
机构：1浙江科技学院，土木与建筑工程学院，中国杭州，310023；2浙大城市学院，城市基础设施智能化浙江省工程研究中心，中国杭州，310015；3浙江大学，建筑工程学院，中国杭州，310058
目的：随着我国城市地下空间的迅猛发展，盾构法因其具有地质适应性强、速度快、安全可靠等优点，逐步成为了城市地铁建设的主流。盾构在开挖掘进的过程中势必会对四周围岩造成一定的扰动，从而引发地层变形与地基的隆起或沉降，进而对临近建筑物造成影响。然而，由于土体介质的复杂性，国内外对于地表沉降规律的研究尚处于不成熟阶段。因此，开展地铁施工过程中的结构健康监测、预测地铁施工引起的周边结构变形以及总结变形规律可以为地铁施工的安全预警提供技术支持和理论参考，对提高地铁施工安全性有重要意义。
创新点：1.以杭州地铁施工为工程背景，分析临近施工对既有地铁结构变形的影响，并研究监测中的难点和重点；2.基于监测获取的海量数据，建立基于反向传播（BP）神经网络的变形预测模型。
方法：分析BP神经网络的4种结构（即单输入-单隐含层-单输出、多输入-单隐含层-单输出、单输入-双隐含层-单输出和多输入-双隐含层-单输出）对预测性能的影响，并通过实测数据进行验证。
结论：1.本工程基坑影响范围内同时存在地铁车站、附属结构和盾构隧道等不同结构形式的地铁设施，且其受基坑施工影响的程度各不相同；保护区监测工作过程中除了关注各自自身结构变形情况外，还应重点加强车站与附属结构、车站与盾构隧道结构间的差异沉降监测。2.对于单输入-单隐含层-单输出的BP神经网络，其最优隐含层节点数为1；对于多输入-单隐含层-单输出的BP神经网络，其最优输入数为2，隐含层节点数为1；对于单输入-双隐含层-单输出的BP神经网络，其最优隐含层节点数分别为1和1；对于多输入-双隐含层-单输出的BP神经网络，其最优输入数为5，隐含层节点数分别为1和5。3.当确定输入量时，隐含层节点数应不大于输入数量，以避免出现过拟合现象；当输入量为1~3时，最优隐含层节点数为1。
关键词组：
References:
Open peer comments: Debate/Discuss/Question/Opinion
<1>
DOI:
10.1631/FITEE.2200573
CLC number:
TN828.6
Download Full Text:
Downloaded:
3235
Download summary:
<Click Here>Downloaded:
224Clicked:
1370
Cited:
0
On-line Access:
2023-07-03
Received:
2022-11-17
Revision Accepted:
2023-01-06
Crosschecked:
2023-07-03