BP神经网络在航空复合材料敲击检测中的应用

Application of BP Neural Network in Detection of Aviation Composite Materials

随着航空复合材料运用越来越广泛,其本身缺陷造成的事故也愈来愈多。提出一种利用敲击检测和BP神经网络的航空复合材料无损检测方法。首先运用敲击检测采集数据;然后运用平均值法和方差法来对数据进行修正;最后借助MATLAB软件进行BP神经网络数据分析,在训练数据4 000组、测试数据20组时,准确率可达90%。实例验证结果表明,基于BP神经网络的敲击检测方法可以实现航空复合材料缺陷的有效检测。

With the application of aviation composite materials more and more widely, the defects of aviation composite materials bring more and more accidents. This paper presents a new method for nonde- structive testing of aviation composite materials based on percussion detection and BP neural network. Firstly ,the defect data is collected by the knock detection;then, the mean value method and variance method are used to correct the data; finally, with the help of MATLAB software to process 4000 groups of training data and 20 groups of the test data, the accuracy rate is up to 90%. The results show that the method based on BP neural network can effectively detect the damage of composite materials.