CAJ | 학술논문

单位时间内通过混凝土结构的总电荷量是反映混凝土结构氯离子渗透能力的一个重要指标.针对传统经验公式法预测了氯离子总电荷量的不足,建立了一个模糊-神经网络模型以预测氯离子总电荷量.此模型能同时考虑多种因素及其非线性耦合作用,方便易行、通用性好,较之其它神经网络具有运算速度快、不容易陷入局部最优、训练效果好的特点.训练样本训练后的结果表明,该模型具有较高的准确度,能可靠预测氯离子总电荷量,准确评价混凝土结构的抗氯盐侵蚀能力,为实际工程的耐久性设计和评价提供依据.
단위시간내통과혼응토결구적총전하량시반영혼응토결구록리자삼투능력적일개중요지표.침대전통경험공식법예측료록리자총전하량적불족,건립료일개모호-신경망락모형이예측록리자총전하량.차모형능동시고필다충인소급기비선성우합작용,방편역행、통용성호,교지기타신경망락구유운산속도쾌、불용역함입국부최우、훈련효과호적특점.훈련양본훈련후적결과표명,해모형구유교고적준학도,능가고예측록리자총전하량,준학평개혼응토결구적항록염침식능력,위실제공정적내구성설계화평개제공의거.
Chloride permeability of concretes,which is determined through rapid chloride permeability test (RCPT),is an important indicator of the ability of chloride ion penetration in concrete.Because it is inaccurate to be predicated by the traditional empirical formula,a new model,based on neural network and fuzzy theory,has been proposed to estimate chloride permeability of concrete.Many related factors and the non-linear coupling effects of them have been taken into account.This model is simple for use and can be effective on many occasions.Compared with other neural networks,the method is better not only because of its high-speed in calculation but also its refusal of local optimum.After training,the testing results show that it has high accuracy.In short,the model is able to predict chloride permeability of concrete reliably and provide assistance for the estimation of durability in practical project designs.