題目：優化方法在巖土工程中的實踐
時間：2019年11月22日15:30
地點：綜合實驗1號樓208室
報告人：尹振宇 香港理工大學土木與環境工程系副教授
摘要:隨著巖土力學理論的發展，高級的土體本構模型不斷涌現，同時模型參數也呈現增加的趨勢且難以簡單確定。但長期以來，由于受模型選擇和參數識別方法發展的限制，無法有效、及時地將高級理論應用到實際工程中來，一定程度上造成了理論的浪費。本報告介紹了現有優化算法在巖土工程中的應用，歸納和總結了不同優化方法在識別參數以及其他方面的優缺點。以從旁壓試驗和基坑開挖為例對比了五種典型優化算法(遺傳算法GA、模擬退火算法SA、粒子群算法PSO、進化差分算法DE和蜂群算法ABC)在識別摩爾庫倫參數中的表現。詳述了實數編碼遺傳算法和差分進化算法的改進及驗證。應用優化方法討論了如何選擇砂土模擬中所需的本構特征、參數識別中目標試驗的類型、數量及應變水平。以基坑工程為例，介紹了基于優化理論的本構模型選擇和參數識別的工程應用。最后，詳述了多目標進化差分算法的改進及其在本構參數識別和工程中的應用。
尹振宇博士，香港理工大學土木與環境工程系副教授
尹振宇，1997年獲浙江大學建筑工程專業學士學位，2006年獲法國巖土工程專業博士學位。曾先后學習和任職于浙江嘉華建筑設計有限公司、法國南特中央理工大學(Ecole Centrale de Nantes)、同濟大學、英國斯特斯克萊德大學(University of Strathclyde)、芬蘭赫爾辛基工業大學(Helsinki University of Technology)、美國馬薩諸塞大學(University of Massachussettts, Amherst)、上海交通大學、香港理工大學，曾于2011年入選上海市特聘教授“東方學者”。兼任美國土木工程師協會（ASCE）“粒狀材料專業委員會”委員及中國多個土力學專業委員會委員，日本土工（S & F）、加拿大土工（Can.Geot. J.）等期刊編委，主要從事土力學與巖土工程方面的教學和研究，近年來在國際核心刊物上發表SCI論文130余篇，WOS的H指數達32。
尹振宇博士的主要研究方向：（1）土體宏微觀特性及本構關系；（2）巖土工程模型試驗及數值分析；（3）人工智能在巖土工程中的應用。
Title: Practice of Optimisation Theory in Geotechnical Engineering
Abstract:In recent decades, various advanced constitutive models for soils have sprung up. The identification of parameters and selection of constitutive models have become one of the hotspots and difficulties in geotechnical engineering, even the engineers are at a loss. Therefore, the purpose of this presentation is to deal with how to intelligently identify models and their parameters, and how to apply them to engineering practice. To this end, the presentation first systematically introduces the basic principles and application fields of various optimization algorithms, and compares the advantages and disadvantages of several typical optimization algorithms in parameter identification, and then presents how to improve the performance of optimization-based parameter identification taking the genetic algorithm and differential evolution algorithm as examples. Several typical geotechnical problems are selected to illustrate the practice of optimization theory, such as: (1) the development of empirical correlations for key mechanical parameters of soils; (2) the parameter identification of natural soft structured clay and granular materials with proposing the least number of objective tests and the type of objective tests; (3) the model selection among various advanced models and critical state related formula; (4) the improvement of multi-objective differential evolution algorithm with its application in updating of predictions during the engineering construction.
Dr Zhen-Yu YIN,Associate Professorof Geotechnics at The Hong Kong Polytechnic University.
Dr. Yin was graduated from Zhejiang University in 1997 for BSc, and then worked as engineer at Zhejiang Jiahua Architecture Design Institute for 5 years. He obtained MSc and PhD in Geotechnical Engineering at Ecole Centrale de Nantes in France in 2003 and 2006.
After PhD, he worked as research and teaching fellow mainly at Helsinki University of Technology (Finland), University of Strathclyde (UK), University of Massachusetts (Umass-Amherst, USA), Shanghai Jiao Tong University, Tongji University and Ecole Centrale de Nantes in France, and HK PolyU.
Dr. Yin’s research interests include: (1) constitutive modeling of soils from micro to macro; (2) finite element analysis for geotechnical engineering; (3) application of artificial intelligence in geotechnics. Since 2008, he has published over 130 SCI articles in peer reviewed international journals with H-index of WOS of 32.
Dr. Yin leads rresearch programmes funded by NSFC, ANR, EU-PF7, RGC etc.He served as member of granular materials committee of American Society of Civil Engineers since 2012. He is currently Editorial Board Member of “Canadian Geotechnical Journal”, “Soils and Foundations”.
