机构地区: 北京航空航天大学
出 处: 《玻璃钢/复合材料》 2006年第5期3-5,31,共4页
摘 要: 研究目的是建立基于人工神经网络的复合材料固化变形预测模型。复合材料固化变形的多因素性致使很难得到精确的解析解。应用人工神经网络方法结合实验实测数据,模拟复合材料各项参数与变形间的非线性关系,对相同材料(玻璃钢)在相近固化条件下的固化变形进行预测,计算速度快,精度高,为固化变形的预测控制提供了一种新方法。 The objective of the research is to set up the curing deformation prediction model of composites based on artificial neural network. It is very difficult to find an accurate resolution to the composite curing deformation from various causations. Three-layer BP neural network is employed to simulate the nonlinear relationship between the curing deformation and causations. Experimental data are applied to train the network as samples. Finally, the trained network is put to predict the curing deformation of composite shells with different thickness sections. The curing deformation of the same material at similar curing conditions can be yielded from the output of BP. The prediction results show that this method is an accurate way at a reasonable computational cost. Further, the method can be applied to control the deformation in curing process.