论文部分内容阅读
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 more and more widespread use of aviation composite materials, more and more accidents caused by its own defects. A method of non-destructive testing of aviation composite material using percussive detection and BP neural network is proposed. First of all, the data were collected by percussive detection. Then the data were corrected by averaging method and variance method. Finally, the BP neural network data was analyzed by using MATLAB software. The accuracy was up to 4 000 training data and 20 test data 90%. The experimental results show that the percussive detection method based on BP neural network can effectively detect aerospace composite defects.