题 目: Distributed Computing with CPU or GPU in the Condition Assessment of a Structure
报告人： Professor S S Law, The Hong Kong Polytechnic University, Hong Kong
主持人： 马宏伟/余岭 教授
时 间： 10:30 AM，2012年5月15日（星期二）
地 点： 蒙民伟理工楼615教室
Synopsis
ALL existing algorithms developed for the condition assessment of a structure are implemented with a single CPU computation. This imposes a minimum period of time for the analysis and it may be up to a few days for the analysis with a large structure. This limitation discourages the use of most existing algorithms for a practical structure which demands long computation time but with a high accuracy.
This seminar addresses this issue by performing the condition assessment of a structure under unknown excitations with the direct analysis of Pattern Search Method. The structure is divided into different zones for short term test in each zone. Measured accelerations from the structure and the finite element model of the structure are taken as the needed information. The unknown excitations are identified in the first stage using CPUs in a multi-core processor. The structure is assessed on its structural conditions with analysis of the structure using sets of measured data from different tests using the GPUs. Results with accuracy similar to that obtained from using single CPU throughout the process could be obtained while there is a significant improvement of 34% in the required computation time.
The idea of parallel computing is further studied by dividing the structural parameters of a structure into several subsets according to the zone divisions, and each zone is test separately with its own set of sensors. The unknown excitations in each zone are identified in an inverse analysis in the first stage in the wavelet domain while the structural parameters are updated inversely in the second stage with the Newton-SOR method with a sub-optimal solution. Each set of excitation and each subset of parameters are updated in separate CPU and the improvement with this decentralized damage detection technique in computation time with a plane frame structure is around 37%.
Short Biography of Prof. S S Law:
Professor Siu-Seong Law, B.Sc., M.Ph., Ph.D.; MHKIE, MICE, MIStructE; MASCE, C. Eng.
Prof SS Law graduated with B.Sc (first class honours) and M.Ph. degrees in Civil Engineering from the University of Hong Kong and gained his Ph.D degree from the University of Bristol, U. K. He joined the Hong Kong Polytechnic in 1979, after working some time with a consulting firm, Maunsell Consultant Asia Limited, Paul Y Construction Co. Ltd, Peter Y S Pun & Associates and Ove Arup Partnership (London) since 1972. Prof Law is now a full professor in structural engineering in the Hong Kong Polytechnic University.
Prof SS Law is an international leading professor working in the field of condition assessment of structures and moving load identification based on measured information of a structure. He has completed the supervision of more than 10 PhD thesis (including 2 PhD thesis jointly with Chinese Universities) and 3 post-doctoral fellows. He is at present a Chair Professor with Beijing Jiaotong University of China.
He has been actively involved in carrying out research on general structural dynamics, inverse problems of condition and system identification, static and dynamic testing, finite element modeling, vibration mitigation, structural health monitoring, bridge structures and so on. Prof Law has published 3 books and 131 SCI journal papers with 44 published in the last five years. Up to March 2012, a total of 136 publications have been included in the SCI-expanded index with the H-index equals 20, SCIE citation number beyond 1000 and the average citation per item of 9.42.
Prof. Siu-Seong Law简介
Siu-Seong Law教授是“桥梁移动荷载”与“结构健康监测”两个领域国际知名学者，更是这两个领域最主要的领军学者之一。他早年从香港大学获得土木工程一等荣誉学士学位和硕士学位，并从英国布里斯托大学获得土木工程博士学位。1979年加入香港理工学院工作至今，现在是香港理工大学结构工程专业的全职教授。Law教授的学术论著包括210余篇学术论文（其中136篇为SCI期刊论文）和3本专著。根据2012年3月Web of Science数据库的统计, Law教授的SCI期刊论文的H-index为20，被他人引用的次数已超过1000次，平均单篇被引次数达9.42。他7次获得香港RGC项目（其中4次连续资助），3次获得港府博士后研究计划资助，多次获得竞争性外部资助项目等，与国内外高校、科研及政府部门，特别是与华南地区的交通主管部门项目合作密切，是中山大学、华南理工大学、北京交通大学等著名高校的客座或讲座教授。
欢迎广大师生踊跃参加！