当前位置:首页 > 报告详情

DC25_SLIDES_Drive_PredictionOfDielectricConstantAnd_Hernandez.pdf

上传人: S** 编号:1240807 2026-05-16 67页 2.72MB

1、Information Classification:GeneralWelcome to ConferenceJanuary 2830,2025 Santa Clara Convention Center1ExpoJanuary 2930,2025 Information Classification:GeneralPrediction of Dielectric Constant&Copper Roughness Parameters of High-Speed Automotive PCB Digital Interconnects Using Data-Based ModelsJose

2、Enrique Hernandez-Bonilla(Robert Bosch GmbH)Golzar Alavi(Robert Bosch GmbH)Torsten Reuschel(University of New Brunswick)Cheng Yang(Hamburg University of Technology)Christian Schuster(Hamburg University of Technology)2Information Classification:GeneralImageSPEAKERJose Enrique Hernandez-BonillaPhD Stu

3、dent,Robert Bosch GmbHJos Enrique Hernndez-Bonilla received his B.Sc.degree in Electronics Engineering from Costa Rica Institute of Technology(TEC),in 2018,and his M.Sc.degree in Microelectronics and Microsystems from the Hamburg University of Technology(TUHH)in 2022.He is pursuing the Ph.D.degree a

4、t Robert Bosch GmbH.His focus is on the measurement and modeling of high-speed PCBs used in automotive applications.3Information Classification:GeneralImage Co-AuthorsGolzar AlaviSenior Manager,Robert Bosch GmbHCheng YangSenior Engineer,Hamburg University of Technologycheng.yangtuhh.de4ImageTorsten

5、ReuschelAssistant Professor,University of New Brunswicktorsten.reuschelunb.caChristian SchusterFull Professor,Hamburg University of Technologyschustertuhh.deInformation Classification:General This research is the result of a cooperation between Robert Bosch GmbH and the Hamburg University of Technol

6、ogy(TUHH)which focuses on the design and characterization of high-speed interconnects for automotive applications.Academia-Industry Context5Information Classification:GeneralAgenda6 Motivationo The Electrification of Vehicles o Automotive High-Speed Interconnectso Automotive PCB Design Challenges Au

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
1. **研究背景**:汽车电子控制单元(ECU)数量激增(约150个/车),高速PCB互连设计面临成本、可靠性与性能的平衡挑战,需精确建模介电常数(εr)和铜粗糙度(Rz)参数。 2. **方法**:采用数据驱动模型(DNN和1D-CNN),基于RLGC传输线模型,通过随机生成的几何与材料参数数据集(εr、tanδ、Rz等)训练,预测Djordjevic-Sarkar介电模型和Cannonball-Huray粗糙度模型的参数(ε∞、Δε、Rz)。 3. **结果**:DNN在单端差分线中表现更优,预测误差低至0.32%(ε∞)、1.06%(Δε)和0.32%(Rz);CNN在差分线中仅需200样本即可达1.60% Rz误差。 4. **应用**:预测参数用于RLGC模型,衰减(α)和相位常数(β)误差均小于5%,验证了数据驱动方法在减少材料表征工作量上的有效性。
**汽车PCB设计挑战** 现代汽车PCB设计面临哪些核心挑战?如何平衡成本、可靠性和性能? **数据驱动预测模型** 如何利用神经网络预测高速PCB的材料参数?这种方法的优势是什么? **环境对PCB的影响** 温度和湿度如何影响汽车高速互连性能?如何通过模型优化应对?
客服
商务合作
小程序
服务号
折叠