计算智能在微尺度力学研究中的应用.pdfVIP

第 26 卷第 6 期 Vol.26 No.6 工 程 力 学 2009 年 6 月 June 2009 ENGINEERING MECHANICS 10 文章编号：1000-4750(2009)06-0010-06 计算智能在微尺度力学研究中的应用 * 1 2 舒 阳 ，舒思齐 (1. 天津大学建筑工程学院，天津 300072 ；2. 中国科学院力学研究所，北京 100080) 摘 要：微尺度力学的研究常遇到力学模型分析、复杂本构方程求解和跨尺度计算等问题，而传统的数学方法很 难给出所研究的微尺度力学问题的解析解。该文探讨运用计算智能方法(主要包括人工神经网络、遗传算法和模糊 数学等知识)对微尺度材料的力学行为进行研究。首先，采用人工神经网络建立微尺度材料的纳米压痕硬度随压痕 深度变化的力学模型，并用其预测氧化镁材料的纳米压痕硬度；其次，运用遗传算法对 A533-B 号钢的球形压痕 的载荷-位移曲线进行反分析，进而获取其力学参数。 关键词：尺度效应；微尺度力学；计算智能；纳米压痕；反问题 中图分类号：O34 文献标识码：A THE APPLICATION OF COMPUTING INTELLIGENCE ON THE RESEARCH OF MICRO-SCALE MECHANICS * 1 2 SHU Yang , SHU Si-qi (1. Construction Engineering College, Tianjin University, Tianjin 300072, China; 2. Institute of Mechanics, Chinese Academy of Sciences, Beijing 100080, China) Abstract: Based on the traditional mathematical methods, the analysis solutions for micro-scale mechanics problems are not usually solved properly for the building of a mechanical model, the solving of comprehensive constitutive equations and multi-scale computing need to be dealt with. In the present paper, computing intelligence, which includes artificial neural network, genetic algorithm, fuzzy mathematics and others, is used to study the mechanical behaviors of micro-scale materials. First, the mechanical model built with artificial neural network, which describes the variety of hardness to the indentation depth, is used to predict the nanoindentation hardness of MgO. Second, the load-depth curves of A533-B steel of indentation