辽宁工程技术大学学报(自然科学版)
遼寧工程技術大學學報(自然科學版)
료녕공정기술대학학보(자연과학판)
JOURNAL OF LIAONING TECHNICAL UNIVERSITY NATURAL SCIENCE EDITION
2013年
1期
115-118
,共4页
桥梁评估%耐久性%模糊数学%神经网络%Matlab 工具箱%层析分析%隶属函数%神经元
橋樑評估%耐久性%模糊數學%神經網絡%Matlab 工具箱%層析分析%隸屬函數%神經元
교량평고%내구성%모호수학%신경망락%Matlab 공구상%층석분석%대속함수%신경원
bridge evaluation%durability%fuzzy mathematics%neural network%Matlab toolbox%analytic hierarchy process%membership function%neuron
将层次分析等理论更快速方便地应用于结构评估工程实践,研究基于现有理论的自动化、智能评估方法.借助 Matlab 模糊数学、神经网络工具箱,采用线性映射的层次分析方法完成桥梁耐久性等级评估. FIS 结构和 BP模型评估结果的对比表明:受隶属函数限制,FIS 结构评估结果较粗糙,但由于模糊关系的特性,处理复杂输入输出关系时效果较好;而当样本数据集在超空间中线性可分时,神经网络对已有数据的拟合精度更高,但只能处理与训练样本同构的评估数据.由此证明:利用 Matlab 软计算模型实现桥梁工作性能的自动化智能评估是可行的.
將層次分析等理論更快速方便地應用于結構評估工程實踐,研究基于現有理論的自動化、智能評估方法.藉助 Matlab 模糊數學、神經網絡工具箱,採用線性映射的層次分析方法完成橋樑耐久性等級評估. FIS 結構和 BP模型評估結果的對比錶明:受隸屬函數限製,FIS 結構評估結果較粗糙,但由于模糊關繫的特性,處理複雜輸入輸齣關繫時效果較好;而噹樣本數據集在超空間中線性可分時,神經網絡對已有數據的擬閤精度更高,但隻能處理與訓練樣本同構的評估數據.由此證明:利用 Matlab 軟計算模型實現橋樑工作性能的自動化智能評估是可行的.
장층차분석등이론경쾌속방편지응용우결구평고공정실천,연구기우현유이론적자동화、지능평고방법.차조 Matlab 모호수학、신경망락공구상,채용선성영사적층차분석방법완성교량내구성등급평고. FIS 결구화 BP모형평고결과적대비표명:수대속함수한제,FIS 결구평고결과교조조,단유우모호관계적특성,처리복잡수입수출관계시효과교호;이당양본수거집재초공간중선성가분시,신경망락대이유수거적의합정도경고,단지능처리여훈련양본동구적평고수거.유차증명:이용 Matlab 연계산모형실현교량공작성능적자동화지능평고시가행적.
@@@@In order to conveniently apply the hierarchical analysis theory in structure evaluation, this study investigates the automatic and intelligent assessment methods based on existing theories. Using Matlab fuzzy mathematics and mathematical toolbox in neural network, the study conducts an evaluation on bridge’s durability with the method of linear mapping and AHP-FCE. Comparative analysis on FIS structure and BP model shows that due to the restriction by membership function, the evaluation results of FIS structure are not accurate, however, because of the characteristics of fuzzy relation, the FIS structure can better adapt to a more complex input/output mapping. When the sample data set can be linearly categorized in super space, the fitting accuracy of existing data using BP neural network is better, however, it can only process the evaluation data which are homogeneous with training sample data. The study proves that it is feasible to use soft calculation model of Matlab to automatically and intelligently evaluate the bridge work performance.