\r郝柏森^{1, 2}，梅云辉^{1, 2}，李 欣^{1, 2}，陆国权\r^{1, 2}\r
\r
AuthorsHTML:\r郝柏森^{1, 2}，梅云辉^{1, 2}，李 欣^{1, 2}，陆国权\r^{1, 2}\r
\r
AuthorsListE:\rHao Baisen^{1, 2}，Mei Yunhui^{1, 2}，Li Xin^{1, 2}，Lu Guoquan\r^{1, 2}\r
\r
AuthorsHTMLE:\rHao Baisen^{1, 2}，Mei Yunhui^{1, 2}，Li Xin^{1, 2}，Lu Guoquan\r^{1, 2}\r
\r
Unit:\r1. 天津大学先进陶瓷与加工技术教育部重点实验室，天津 300350；
2. 天津大学材料科学与工程学院，天津 300350\r
\r
Unit_EngLish:\r1. Key Laboratory of Advanced Ceramics and Machining Technology of Ministry of Education，Tianjin University，Tianjin 300350，China；
2. School of Materials Science and Engineering，Tianjin University，Tianjin 300350，China\r
\r
Abstract_Chinese:\r模块的布局设计是功率模块设计的关键环节，随着功率模块向着集成化、高频化飞速发展，布局设计对于模块可靠性的影响越来越重要．目前，功率模块的布局设计主要依赖于设计人员经验而以手工方法完成，在此过程中消耗了大量的人力和时间成本．电子设计自动化作为未来电子领域的发展趋势，近年来已经研究用于功率模块的设计之中．然而，传统遗传算法的应用造成自动化方法无法在可接受时间内得到准确的结果．针对该现状，设计开发了结果精度更高、运算速度更快的修正自动化布局遗传算法．该算法采用统计学、种群多阶段思想与遗传算法思想减少种群数量与运算复杂度，最终保证算法计算速度和精度的提高．将该算法用于简单的半桥模块实例的布局设计，并采用ANSYS Q3D 软件验证，结果表明，相比于原来的自动化布局设计遗传算法，本文方法缩短了约29%的运行时间，并且布局方案的适应度平均值达到231.14，其标准差仅为0.43，优于原有方法175.35 的适应度平均值和13.96 的标准差．所以，本文方法可以在短时间内得到更准确的结果，这有助于将自动化方法真正用于实际功率模块的布局设计．\r
\r
Abstract_English:\rFloorplanning is one of the critical steps in designing a power module. With the rapid development of power modules toward high integration and frequency，floorplanning is becoming increasingly significant in ensuring reliability of power modules. Presently，the layout of a power module is mainly designed manually based on the designer’s experience，which is labor-and time-consuming，whereas，electronic design automation，as a development trend in the field of electronics，has been studied in recent years for the design of power modules. However，the application of traditional genetic algorithm cannot realize accurate layout results in a short time. To increase the calculation accuracy and speed，a modified genetic algorithm was developed in this study. By applying the concepts of statistics，multi-stage population，and genetic algorithm in the modified genetic algorithm，the population and complexity of operations were significantly reduced，which resulted in higher calculation accuracy and speed. Then this method was used for layout design of a simple half-bridge module and verified by ANSYS Q3D software. Compared with the traditional genetic algorithm，the modified genetic algorithm method saves calculation time by about 29%. Additionally，average fitness was increased from 175.35 to 231.14，and the standard deviation was reduced from 13.96 to 0.43. Therefore，the proposed modified genetic algorithm method can realize more accurate results within a short time，and this will contribute in making automated layout designs for power modules practicable.\r
\r
Keyword_Chinese:功率模块；自动化布局；多阶段遗传算法；寄生参数；低寄生电感\r

Keywords_English:power module；automated layout design；multistage genetic algorithm；parasitic parameters；low parasitic inductance\r


PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6244