CAJ | 학술논문

对超弹性形状记忆合金（SMA）丝在不同应变幅值和荷载速率下进行加卸载单轴拉伸试验,分析其滞回特性随环境因素的变化规律。将径向基函数神经网络（RBFNN）和Graesser模型结合起来,Graesser模型参数取自试验曲线,能由数学式确定的模型参数和应变幅值、荷载速率一起作为网络的输入信息,不能由数学式确定的模型参数作为输出神经元。数值计算表明,RBFNN可以精确地预测Graesser模型参数,且计算的SMA应力-应变曲线与Graesser模型结果吻合很好。
대초탄성형상기억합금（SMA）사재불동응변폭치화하재속솔하진행가사재단축랍신시험,분석기체회특성수배경인소적변화규률。장경향기함수신경망락（RBFNN）화Graesser모형결합기래,Graesser모형삼수취자시험곡선,능유수학식학정적모형삼수화응변폭치、하재속솔일기작위망락적수입신식,불능유수학식학정적모형삼수작위수출신경원。수치계산표명,RBFNN가이정학지예측Graesser모형삼수,차계산적SMA응력-응변곡선여Graesser모형결과문합흔호。
One-dimensional loading-unloading tests on superelastic SMA wires were performed at varying strain amplitudes and loading rates to evaluate the effects of strain amplitude and loading rate on the hysteretic behaviors.The combination of the Graesser＇s model and the radial basis function neural network（RBFNN） was proposed,i.e.the parameters of the Graesser＇s model were acquired from the experimental data,the parameters that were determined by the mathematical expressions,strain amplitude and loading rate constitute the input information of the network,and the output neurons were made up of the Graesser＇s model parameters that were not determined mathematically.Numerical simulations indicate that the RBFNN can predict the parameters of Graesser＇s model accurately,and the simulated stress-strain curvesby the RBFNN-Graesser＇s model agree with the results of the Graesser＇s model very well.