DEVELOPMENT OF PROGRAM ON NONLINEAR STRUCFURAL ANALYSIS USING NEURAL NETWORK

在建立非弹性材料和构件的弹性力学行为的弹簧-质量模型时，通常考虑的是基于荷载-变形试验结果的数学模型。该模型必须近似于材料的非弹性滞后。然而，使用数学模型对材料的行为进行假设是至关重要的，因为如果针对特定情况应用不适当的模型，可能会导致严重的错误。本研究以多层式类神经网路模拟非线性滞回特性，如Ramberg-Osgood模型、修正双线性模型及Takeda模型。本文利用神经网络的模式识别能力，直接用神经网络对非线性滞回性能进行建模，而不用数学模型代替。评价了该网络在数值分析中的有效性和适用性。构造了用于非线性滞回曲线评价的广义多层神经网络。该网络可以很好地识别三种类型的滞回曲线。该网络可作为非线性弹簧动力响应分析的子程序。

Generally, in formulating a spring-mass model for hysteric behavior of materials and members with inelastic characteristic, a mathematical model based on load-deformation experimental results is considered. The model must approximate the inelastic hysteresis of the material. However, assumption of material's behavior using mathematical models is crucial, since it may cause serious errors if inappropriate model is applied for a particular situation. This research describes multiple layered neural network to simulate the non-linear hysteretic behavior like Ramberg-Osgood model, modified bilinear model and Takeda model. In this study, based on the pattern recognition ability of neural network, non-linear hysteretic behavior was modeled by the network directly without replacing it with a mathematical model. The effectiveness and applicability of the network in numerical analysis were evaluated. Generalized multiple layered neural network to evaluate non-linear hysteretic curve was constructed. The network can recognize well the three types of hysteretic curve. The network is available as a subroutine of non-linear spring in dynamic response analysis.