结合小波分析和BP神经网络的复合材料损伤检测技术研究

发布时间：2018-08-17 15:43

【摘要】：复合材料具有质量轻、强度高、耐腐蚀等优点,已经广泛应用于军事、航天、交通、电子电气等领域。由于复合材料很容易遭受外来破坏而产生损伤,所以对复合材料进行损伤检测就显得很重要,目前用光纤光栅传感器对复合材料进行损伤检测是研究热点之一,本文将开展光纤光栅传感器采集信号,结合小波分析和神经网络的信号处理方法,实现复合材料的损伤检测。本文先介绍了复合材料的研究背景和意义,然后分析了关于复合材料损伤检测国内外的研究现状,同时分析了小波分析和神经网络在损伤检测方面的应用。接着介绍了小波变换的基本理论,研究了一种新的Lab VIEW和MATLAB混合编程的信号解调技术,实现信号实时处理。结合复合材料板冲击响应信号的特点,研究了信号的小波包能量谱分析方法,能够提取到信号不同频率上的特征信息,为提取损伤特征向量提供理论了依据。随后进行复合材料的冲击实验,研究了传感器布设与响应信号之间的关系,结果表明当传感器的轴线与冲击点和传感器中心连线垂直且距离冲击点较近布设时传感效果最好,之后在复合材料板上模拟损伤得到不同损伤状况时的冲击响应信号,通过小波包能量谱分析得到了信号的损伤特征向量。最后,结合从四个传感信号提取的损伤特征值和复合材料板不同的损伤工况,组成BP神经网的学习样本,提出了通过小波包分析得到信号能量谱提取损伤特征向量,输入BP神经网络进行损伤识别的方法,构建了一套结合小波分析和神经网络的复合材料损伤检测系统,进行复合材料的损伤检测。
[Abstract]:Because of its advantages of light weight, high strength and corrosion resistance, composite material has been widely used in military, aerospace, transportation, electronic and electrical fields. The damage detection of composite material is very important because it is easy to be damaged by external damage. At present, it is one of the research hotspots to use fiber Bragg grating sensor to detect the damage of composite material. In this paper, the fiber Bragg grating sensor will be used to detect the damage of composite materials by combining wavelet analysis and neural network signal processing. This paper first introduces the research background and significance of composite materials, then analyzes the research status of composite damage detection at home and abroad, at the same time, analyzes the application of wavelet analysis and neural network in damage detection. Then, the basic theory of wavelet transform is introduced, and a new signal demodulation technique of mixed programming of Lab VIEW and MATLAB is studied to realize the real-time signal processing. According to the characteristics of composite plate impulse response signal, the wavelet packet energy spectrum analysis method of the signal is studied, which can extract the characteristic information of the signal at different frequencies and provide the theoretical basis for extracting the damage characteristic vector. The relationship between sensor placement and response signal is studied. The results show that the sensor is the best when the axis of the sensor is perpendicular to the impact point and the center of the sensor is connected and the sensor is close to the impact point. Then the shock response signals of different damage conditions are obtained by simulating the damage on the composite plate, and the damage characteristic vectors of the signals are obtained by wavelet packet energy spectrum analysis. Finally, according to the damage characteristic values extracted from four sensing signals and the different damage conditions of composite plates, a BP neural network learning sample is formed, and the signal energy spectrum is proposed to extract the damage feature vector by wavelet packet analysis. The method of BP neural network is inputted to identify the damage, and a composite damage detection system combining wavelet analysis and neural network is constructed to detect the damage of composite material.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TB33;TP18

【参考文献】