摘要
创建了基于BP神经网络方法预测复合材料在复杂应力状态下的失效模型。依据现有的试验数据,建立了基于BP神经网络的数学模型,并给出了网络在多次训练的统计性评价参数,以及单次样本泛化性能的评价参数,并利用这些参数实现了网络的结构优化和训练优化,得到了比较精确的网络模型,成功预测了复合材料的失效数据,拓宽了计算复合材料失效的范围,同时可以对复合材料的失效分析和试验设计进行指导。
The model based on back propagation neutral network （BP） is constructed which is used to predict the failure of composites under combined load. Depending on existing experimental data, mathematical model is built, and the estimation parameters for both statistical and single data sample are introduced to evaluate the performance of BP neural network, at the same time, we utilize those parameters to complete structural optimization and training optimization, which promote the accuracy of the network. The experience of BP neural network gives a new approach to predict the failure of the composites, and guides the designer for better understanding the materials and makes them more confidence to use composites while designing new products.
出处
《航空计算技术》
2009年第2期29-32,共4页
Aeronautical Computing Technique
关键词
复合材料
BP神经网络
失效分析
结构优化
训练优化
composites
BP neural network
failure analysis
structural optimization
training optimization
