摘要
三维编织复合材料由于其材料结构及编织工艺的复杂性和众多工艺参数的影响，目前尚未建立成熟的力学模型。本文采用人工神经网络ＢＰ（ｂａｃｋｐｒｏｐａｇａｔｉｏｎ）算法，将编织工艺参数作为人工神经网络的输入，将弹性模量及强度性能作为输出，建立了编织工艺参数与力学性能的人工神经网络关系模型，并讨论了ＢＰ算法及网络构造。这种人工神经网络关系模型对于三维编织复合材料的实验、生产和应用，工艺参数的选取以及理论模型的研究都有重要的参考价值。本文最后给出了三维编织复合材料拉伸和压缩性能的实验结果与人工神经网络预测结果，通过对比显示，其模拟效果令人满意。
As the complexity of braided technique and effects of various technological parameters, a reasonable mechanical model of 3D braided composites has not been developed yet. This paper presents an artificial neural network relation model with technological parameters and the mechanical behavior of 3D braided composites using BP algorithm. These parameters are used as inputs, the elastic module and the strength are used as outputs. The BP(back propagation) algorithm and the network construction are also discussed. The neural network relation model is an important reference to the experiment, manufacture, application of 3D braided composites and selection of technological parameters for studying theoretical model. Additionally, the paper provides a contrast between the tension/compression test results and the simulating results by the artificial neural network. It is shown that the simulation is successful.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
1997年第4期397-401,共5页
Journal of Nanjing University of Aeronautics & Astronautics
基金
航空科学基金
关键词
神经网络
纤维增强
复合材料
材料力学性质
fiber properties
neural networks
fiber reinforced composites
material mechanical property, 3D braided composites
