基于BP神经网络的SiC_P/AZ61复合材料力学性能预测

Prediction of mechanical properties of SiC_p/AZ61 composites based on BP artificial neutral network

利用人工神经网络(ANN)的BP(back propagation)算法,建立了复合材料力学性能预测模型。模型由三层神经元组成,分别为输入层、隐含层和输出层。以SiC_P/AZ61复合材料的力学性能与SiC的颗粒体积分数的关系为研究对象,选取了七组试验数据作为学习样本,用建立的网络预测未知,并给出预报曲线和预测程序界面图。与试验结果比较表明,所建立的网络能反映SiC_P/AZ61复合材料中SiC的颗粒体积分数与其力学性能之间的关系,为试验设计提供了一种新的思路。

A model is established to predict mechanical properties of the composites with the back propagation （ BP ） algorithm artificial neural network （ANN）. It is composed of three neuron layers： input layer, hidden layer and output layer. The relationship of mechanical properties of SiCp/AZ61 composites and SiC particle volume fraction is researched, 7 samples are used as study data, and the unknown data can be predicted by the model and the prediction curve and the interface of prediction proceeding are drawn. Compared with the experiment, it shows that the established network can reflect the relationship of SiC particle volume fraction of SiCp/AZrl composites and mechanical properties, which provides a new route for experiment design.