Research On Structure Defect Detection Based On Vibration Signal

Structure defect detection is an interdisciplinary and practical technology evaluating the conformity of mechanical components in shipbuilding,aerospace engineering and civil construction.Traditional nondestructive testing methods were expensive and time consuming,thus,precluding industrial automation.The development of modal theory,modern testing and signal processing has jointly promoted structure detection techniques based on vibration and sound signals.How to use the modal parameters to detect structures automatically in industrial setting is worthy of further study.This paper presented the researches on a batch of ceramic tiles of various typesand with various defectives,to investigate the ideal detection method for industrial settings.The main contributionsand conclusions are summarized as follow:1.The theoretical modal analysis was conducted,to analyze the sensitivity of natural frequency and vibration mode against different geometry,size,density,backside texture and other physical parameters.The results suggested that the two modal parameters were sensitive with excitation frequency below 3 kHz.Compared to vibration mode,the natural frequency was more sensitive over a wide range of frequencies.2.The experimental modal analysis was conducted to identify the proper modal parameters.The baselines for tile samples were determined first by comparing experimental and theoretical results.Among these modal parameters,natural frequency,damping ratio and vibration mode were found not feasible while frequency response function was more suitable under automated industrial testing as it encoded the information of natural frequency and vibration mode.3.Various similarity algorithms according to the characteristics of the frequency response curve were studied.An improved normalized Euclidean distance and curves correlation were proposed for the detection.This method was proved effective and not sensitive to the additional mass of the accelerometer.4.Defect detection using acoustic signal was explored and discussed.Natural frequency can be identified,and the similarity of the acoustic frequency response curve can be used to distinguish the adhesion,crack and other larger defects.