用神经网络模拟混凝土的破坏准则

《用神经网络模拟混凝土的破坏准则》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、摘要拳

2、料的徽观细观结擒决宠材料的缴戏性虢。要从材料的微观缨潞缭构层次怎分析糖瓣懿鬣分、维梅毒宏鬟畿辘豹关系，醛缓导箍—含戆蒉雾苓瓣结祷蒺捩熬逶怒携灌蠹。实际t材科内部结构豹澎成发展及破林建程，就是髓爨转化静过稚。黥量就蕊这榉一令能够贯穿不同层次的通用物理基。瀑凝未鞭袈蕤是赣究灌凝主霰筇、辩裂韵～巾蹙簧豢标。论交黄燕奔绥7游凝±戆畿爱蕤，论述了影豌灌凝±蘩袋蕤豹冬今嚣豢，鑫及嚣垂涎凝±瓣襞麓翡冬释方法。建立了糖经瘸络搂墼，糖够麓麓度瓣预测漫凝土戆精裂能。断裂面是材料断裂尉留下的真实记录和霆要信患，对追溯断裂行为以及探索断裂枫蓬÷分有意义。混凝主凝裂嚣在一定藏疆蠹是鸯

3、很好的壹梗锹黪缎。论文用分彤几何豹蘧谂氍方法撵尊其势澎特摄戆溺试帮译徐，r为豢索漫凝±甄裂瑟静澎特征专蘩装簿藩之间的内_在规律以及为髓艏建立相关的数学模溅提供了蓬要的依据。混凝土的强度来源予兰个努蘑；浓泥砂浆鏊体涟液；程鹭辩瓣颓敉强度(群夤耨羝攘辨宠豹毙力)；戳及窳淀砂浆与粗髂料的鞑结力。本文应用麓像处理技术，对断裂聪避移了_兰次处理，得到了霄焉鼹结论，并送行了探索链鹃骚究。凳滋翳静磺究歼肄了瀵路。随后，论文讨论了研究滟凝土破辫：的几种观点，并采用虢囊的观点研究漓凝土的破环。本论文从文献中搜煞了大量的混凝土破坏数据，应用神缀网络进行训练模j棼【，根据豢蠢、耗戆原理建

4、立了滋凝纛静疆坏接粼。并羹意这个薪静淫凝主破坏准裂器蕊窀鼹逡磐了篦骏。关键溺：瀑凝童；蔽缮灌掰；精餮裴；鼗装瑟鲞穆；精经鼹终；静茨辩建疆冀浚；搂髹；破坏准翔：MATLAB：AbstractThepropertyofmaterialisdeterminedbyit’smicroscopicstructure．Andacertainphy+sicalquantitythatcanpenetratefrommicrocosmtomacrocosmshouldbefoundtoanalysistherelationofmaterialpropertyanditsconsti

5、tuentsandstructurefrommicroscopicview．Energyistherightphysicalquantitythatcanbeusedinbothmicroscopicandmacroscopicview，forthereasonthattheprocessofstructure’sformation，developmentandfailureisexactlyaprocessofenergytransformation。Concretefractureenergyisaphysicalquantitythatisveryuseful

6、inthestudyofconcretefailureandfracture．ThisdissertationfirstlyintrcIducestheconceptofconcretefractureenergy，andConcretefractureenergyisforecastedwiththemodelbasedonBP(Back-propagation)network，whichiswidelyusedinNeuralNetwork．Forecastedvaluesarecomparedwithexperimentdataandthevaluescalc

7、ulatedbyempiricalformulae．TheresearchindicatesthatNeuralNetworkhasahighaccuracybecauseitisabletorealizenonlinearrelationsandbecausemorefactorsthatinfluencefractureenergyaretakenintoaccount．TheNeuralNetworkforecastmodelhasawideapplyingperspectiveintheforecastofconcretefractureenergy．Sin

8、cefracturesurfaceisthetruerecordandinformationthatremainfromfractureofmaterial，ithasimportantvalueinthefractureretroactivityandintheexplorationoffractureprinciple．Concretefracturesurfacehasthecharacteristicoffractalgeometry．Thefracturesurfacewasprocessedbeforedigitalimageistaken．Thed