材料本构模型的唯一性  

Uniqueness of Material Constitutive Model

作　　者：杨成祥[1] 冯夏庭[1] 王泳嘉[1] 

机构地区：[1]东北大学资源与土木工程学院,辽宁沈阳110006

出　　处：《东北大学学报（自然科学版）》2000年第5期566-568,共3页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金! ( 5960 4 0 0 1);教育部博士点基金! ( 960 14513)

摘　　要：利用作者最新提出的材料本构模型智能识别的进化学习算法 ,结合实例分析 ,从一个新的角度对该问题进行了阐述 ,证明了刻意追求学习效果的不合理性·指出根据实验数据建立材料本构模型的正确方法应该是使获得的本构模型不仅对学习样本而且对类似条件下的应力分析都能获得很好的效果·并说明了进化学习算法是解决问题的一个好方法 ,为材料本构模型的研究提供了一个新的有力工具·The material constitutive model is usually obtained by fitting analysis on experimental data. Generally,there are several fitting results,thus,the uniqueness problem occurred. It brings difficulties in selecting the result,and it often led to the misunderstanding that the more accurate the fitting result is,the better the obtained model will be. Using the evolving learning algorithm (ELA),proposed by the author recently,which can be used to perform the intellective identification of material constitutive model,the paper stated the uniqueness from a different point. The result shows that there is no direct mapping relationship between fitted result and predicted result,and it is unreasonable to pursue the fitting result excessively. Pointed out that the correct model obtained form experimental data should works well in the numeric simulation on both the experiment sample and other similar problem. It is showed that the ELA can find the global best answer,it is a good method to solve the uniqueness problem,and provides a new theory for the study of constitutive model.

关 键 词：岩土材料 本构模型 唯一性 进化学习算法 

分 类 号：TU452[建筑科学—岩土工程]