Structural Modal Recognition And Damage Identification Supported By 3D-DIC Vibration Measurements

Engineering structures require accurate modal parameters for damage detection,property evaluation,and dynamic evolution analysis.Compared to traditional contact measurement methods,non-contact Digital Image Correlation(DIC)measurement methods do not introduce any additional mass or external damping to the structure and can achieve precise identification of modal parameters.Therefore,this paper applies DIC measurement methods to structural modal analysis and modal-based dynamic damage identification.It conducts systematic theoretical analysis,numerical simulation,and indoor experiments research around the scientific and technical issues faced in the DIC application process.The main research work and conclusions are as follows:(1)Analyzing the measurement errors in the DIC method for vibration testing and classifying the measurement errors.The paper proposes measures to reduce operational errors and random errors.To reduce operational errors,high-quality random speckle patterns generated by "Glare" software are transferred onto the surface of the structure under test,and the imaging distance when the size of speckle particles is set to 4 pixels is determined.To reduce random errors,the paper introduces the wavelet threshold denoising method for signal denoising preprocessing.The impact vibration measurement experiment of the thin steel beam shows that when selecting 3～5 decomposition scales,adaptive thresholds,and double threshold functions,the denoising effect of the displacement response signal is the best.(2)The paper conducts research on modal analysis methods that conform to DIC measurements and proposes a hybrid modal parameter identification method(referred to as the hybrid method).The vibration test experiment of the cantilever aluminum plate is carried out using the DIC measurement method,and the classical modal analysis method is directly used to extract the modal parameters from the displacement response collected by DIC.The results show that the classical method cannot accurately and efficiently identify all the modes of the cantilever aluminum plate from the low signal-to-noise ratio and short data collection time of the DIC data.The proposed hybrid method has higher modal recognition accuracy and completeness,can efficiently and accurately identify all modes of the cantilever aluminum plate in the range of 0～500Hz,and is different from the classical method.(3)The paper conducts research on the automatic selection method of real poles in the stability diagram and develops modal parameter automatic identification software.Based on a large number of tests,the empirical formula for modal distance is defined,and the DBSCAN density clustering algorithm is introduced to intelligently cluster the scattered stable points.In order to reduce the sensitivity of the DBSCAN algorithm to unreasonable input parameters,the paper combines the three-sigma criterion with excellent outlier detection ability.A real pole automatic selection method is constructed by integrating the DBSCAN algorithm and the three-sigma criterion,and the reliability of the method is verified by numerical simulation and indoor experiment.Finally,to facilitate operation and reduce the use threshold,an automatic modal parameter identification software suitable for DIC measurement is developed based on the App Designer platform of MATLAB.(4)The research on modal-based dynamic damage identification proposes a damage identification method for structures without a baseline,which has high positioning accuracy and robustness to noise.By conducting two-dimensional continuous wavelet transform on the obtained modal shapes,the corresponding damage identification indicators are constructed.ANSYS is used to establish finite element models of cantilever plates with breathing cracks and rectangular notch damage,respectively,to verify the feasibility and effectiveness of the proposed method from the perspective of numerical simulation.The numerical simulation results show that the proposed method has the ability to highlight damage and suppress noise,and can accurately identify small damage at a level of 10% in noisy(even high-noise)environments without additional noise reduction algorithms.Finally,the damage identification experiment of a scratch-damaged plate is conducted to verify the practical applicability of the proposed method.