Study On Damage Monitoring Of Composite Materials Based On Fiber Optic Sensor And EMD

Composite material have a lots of advantages, such as light weight, high strength, wear resistance and corrosion resistance etc,It has been widely used in the military, aerospace, medical, chemical, construction, automotive, marine and other fields. However, due to low sensitivity of damage, composite material's strength and carrying capacity is easily to be decreased, so the real-time monitoring for the damages of composite material is very necessary in some applications. Because of the anistropic of the composite material, the signal that the monitoring system collected is non-linear,non-stationary,how to handle these data is the difficult in the monitoring of composite materials.The FBG sensor as a new type of sensor,with high stability, anti-interference, easy of distribution and other advantages, has been used in various fields of engineering and also has a good performance. EMD is an adaptive decomposition based on the local time-varying characteristic of the signal, it can decomposed the signals into numbers of IMFs and a residual, it overcomes the limination of using the harmonic components in the traditional method to represent the non-stationary, non-linear signals,and more suitable for processing non-stationary signals.Artifical neural network has strong learning ability and parallel computing ability,especially suitable for dealing with nonlinear problems.Based on this,this paper has combined the FBG sensor with EMD decomposition and neural network technology to monitor the damage of composite material.The main work in this thesis includes:1. analyzed the the problems of composite material in practical application. The strain characteristics of fiber bragg grating under the axial and radial stresses are discussed,also the sensitivity of the FBG sensor is obtained.2. Established and analysied the finite element model of the composite board, obtained the law of the strain distribution of composite panel in the case of health and injury.In the same load,the strain under the injury is greater than that in the healthy,also around the damaged area stress is concentration.3. Useing the results of the finite element analysis, the position of the pasting place and the range of the monitor array is determined.Then using the SM130fiber demodulator,the change of wavelength under the same load of different locations in the states of health and injury are collected,the waveform under the same energy impact of different locations in the two states are collected.4. Using the EMD to processing the data,the IMF components of each signal are obtained, proposed a method of vector extraction based on the IMF energy contribution rate. Established a three-layer BP neural network with12neurons in hidden layer,and using the feature vector to train the network,similar characteristics of the input sample testing showed that the neural network can be very good for damage identification.5. Using the relationship of the distance between the propagation path and the damage point impact on energy,the positioning of the damaged area is located.