Research On Vibration Modal Testing Method For Thin-walled Components Using Monocular Vision Corner Tracking Approach

Thin-walled components have been widely used in various engineering fields,since it has advantages of light weight, compacted structure and large loading capacity. The efficient and accurate acquisition of vibration modal parameters of the thin-walled components is the precondition and the base to investigate the vibration behaviors analysis and the structural damage identification of the thin-walled components.The accelerometers are used in traditional testing to obtain modal parameters, however the dynamic characteristics of the thin-walled components will be affected by the additional mass of the accelerometer, meanwhile the spatial sensing resolution is low and difficult installation of accelerometers, cannot be solved since the limited number of accelerometer. Single point laser scanning and other optical methods covers the shortage of contact modal testing, but this method requires a complex optical system, the device for controlling and executing, these system and devices need to be complicated aligning and debugging. The machine vision vibration testing method has many advantages, such as full field measurement,non-contact, high precision, convenience and simplicity. It has been widely used in vibration modal testing. Aiming at the existing machine vision vibration testing method obtain the vibration signal through high-contrast markers which are installed on the surface of the equipment in advance, put forward the method that the physical markers replace by virtual corners to get vibration signal. the vibration modal testing method for thin-walled components using monocular vision corner tracking is developed, the main contents are as follows.Firstly, the basis of the vibration modal testing method for thin-walled components using monocular vision corner tracking approach is studied in this paper,the principle and deficiency of the existing vibration modal testing methods base on machine vision are outlined. The in-plane vibration measurement model of thin-walled cmponents under testing based on the image-forming principle of monocular industrial camera is established; The corresponding relationship between edge feature points of thin-walled components in vibration image sequences is analyzed; The camera distortion correction method is studied, the experiment on camera calibration and distortion correction is implemented, and the correction effect is verified according to the collinear feature of corners; the conversion relationship between image pixel and the actual size is obtained based on aforesaid in-plane vibration measurement model; how pixel coordinates of corner obtain frequency information and recognize mode shapes is studied.Secondly, the key steps of the vibration modal testing method for thin-walled components using monocular vision corner tracking approach is further studied, that is, edge detection and corner tracking; an improved Canny edge detection method after analyzing classic edge detection algorithm is proposed to improve the accuracy of edge detection; an improved corner detection method based on Curvature Scale Space (CSS) is proposed in response to the fact that traditional corner detection method can hardly position virtual corners accurately; a rough corner matching method with the constructed corner feature descriptor being the benchmark is introduced, and a matching method based on the quadratic sum of pixel gray difference is applied to conduct precise corner matching, and thereby effective and accurate tracking of virtual corners in vibration image sequence is realized.Finally, the vibration modal testing experiment of thin-walled components is studied, the vibration modal testing scheme of thin-walled components based on monocular vision is proposed, the vibration modal testing system for thin-walled components is built, including industrial camera, optical lens, light source and modal vibration exciter, etc. The sweep frequency testing of the thin-walled beam and thin-walled cylinder is done, the first and second orders natural frequencies of the thin-walled beam and thin-wall cylinder are obtained, the testing results are compared and verified with the results of existing vibration modal testing method base on machine vision and finite element simulation analysis, the results show that this method can accurately measure the modal parameters of thin-walled components without any preinstalled markers,the complicated problem of paste markers is effectively solved, the spatial resolution is greatly improved, it provides a convenient, reliable and accurate method for vibration modal testing of mechanical structures.