Composite Material Damage Detection Research Combining Wavelet Analysis And BP Neural Network

Composite materials have the advantages of light weight, high strength, corrosion resistance and so on, have been widely applied to military,aerospace, transportation, electrical and electronic fields.Due to the composite material is vulnerable to external damage resulting in injury, so the damage detection of composite material is very important. At present,the research of efficient detection method on composite materials is less. In this paper, using fiber Bragg grating sensor to acquisition the signal,combining wavelet analysis and neural network signal processing methods to realize the damage detection of composite material.In the first of the paper introduces the research background and significance of the composite material,then analyzes the research status and development of composite material damage detection at home and abroad. At the same time,studies the applications of wavelet analysis and neural network on damage detection. Then introduces the basic theory of wavelet transform and adopted a new signal demodulating technology of Lab VIEW and MATLAB hybrid programming,realizing real-time signal processing.Then use the wavelet packet energy spectrum analysis method of signal,can extract the characteristic information of signal on different frequency, providing theory basis for extracting damage feature vector.And conducts the composite impact test,researching the relationship between the sensor layout and response signal. The results show that when the axis of sensor perpendicular to cable between the impact point and center of the sensor,and close to the impact point,the sensor works best.And then imitates composite material plate damage to shock response signal of different damage conditions,obtaining the damage characteristic vector by wavelet packet energy spectrum analysis.At last,combining with the damage characteristics value extracted from four sensing signal and different damage conditions of composite material plate constitute the learning samples of BP neural network,put forward a method of utilizing wavelet packet analysis to get the energy spectrum of the signal and extracting the damage feature vector,inputting to the BP neural network to identify the damage, carrys out a set of composite materials damage detection system based on the wavelet analysis and neural network to proceed composite materials damage detection.