Research On Lamb Wave Structure Damage Degree Evaluation Method

It is particularly important to carry out structural monitoring and safety assessment for large scale structures,including identifying the potential safety hazards in a timely manner,and establishing a safety assessment system for the structures.The damage mechanism of composite materials is usually very complicated for many influencing factors.It is difficult to find a clear functional relationship to diagnose the damage and evaluate its degree.The artificial neural network(ANN)potentially can be used to solve this problem due to its non-linear mapping ability.The damage monitoring mechanism was analyzed for composite structures firstly.The characteristic parameters were extracted related to typical damages.ANN technologic was used as an evaluation algorithm for diagnosis of the degree of damage.And different degrees of typical damage were tested by using this evaluation model.The following aspects were mainly studied in this thesis:(1)the theory of structural health monitoring and the propagation of waves in composite materials were introduced,including the dispersion characteristics of Lamb waves,excitation and sensing methods based on the principle of piezoelectricity of PZT element.The basic methods of damage assessment based on Lamb waves were also analyzed.The basic principle of ANN was also explained.(2)Signal processing methods for the responses of structures before and after damage were analyzed in traditional structural health monitoring(SHM)technologies.The wavelet packet method in time-frequency domain analysis method was used to analyze the collected structural responses,especially the ones after structure damaged.Four characteristic parameters were extracted finally: time domain characteristic parameters(waveform Wf,wave peak Wp)and frequency domain characteristic parameters(energy distribution Ed,energy percentage E)at different degrees of damage.Based on the Euclidean distance,the intra-class distance,inter-class distance,and separability criteria were constructed and the damage characteristic parameter vector was established then.(3)A Lamb wave based structure damage assessment model was designed and established.By analyzing and comparing the traditional method of damage degree evaluation methodologies,an ANN with typical three-layer network architecture was selected as the core of the damage degree evaluation.On this basis,a damage assessment model was constructed.(4)A Lamb wave based structural damage assessment system was designed and completed from two levels,named as hardware construction and software design.By using this system,the analysis of the structural responses and the evaluation of the damage degree could be achieved.Finally,experimental verification was carried about on a composite plate,and a linear PZT sensing array was designed and implemented.The system was experimentally verified for typical damages.The damage degree was evaluated by BP ANN and GA-BP ANN respectively.The experimental results showed that the Lamb wave-based damage assessment technology proposed can evaluate different degrees of typical damage well.It could be used to improve the research on damage prediction and provide valuable information to reduce potential safety hazards.In this respect,the method proposed in the thesis indicated potential application in the future.