Research On Real-time Vibration Monitoring Method Of Dynamic Equipment Based On Machine Vision

As a key power export and energy transforming device,dynamic equipment undertakes the mutual conversion between mechanical energy and other forms of energy,and plays an essential role in the fields of national defense construction and industrial production.As the capacity of modern industrial enterprises developing and the scale of production expanding,the types of dynamic equipment are increasing,the mechanical structure is becoming more and more complex,and the vibration problems generated during the operation of equipment are becoming more and more serious,bringing more and more potential safety hazards to the production of enterprises.Vibration monitoring technology of dynamic equipment has become a key research content for enterprise production safety management departments and relevant research institutes.Machine vision-based vibration monitoring technology is a non-contact vibration monitoring technology with video motion amplification method as the core.This method can amplify the sub-pixel level tiny movements that are difficult to be directly observed by human eyes to the extent of visualization by human eyes,such as the slight swing of a bridge under the action of wind load,human pulse,heartbeat,and the tiny trembling of objects struck by external forces.In terms of dynamic equipment vibration monitoring,with the help of machine vision-based dynamic equipment vibration monitoring technology,the tiny vibrations accompanying the operation of equipment can be magnified to the extent that they can be observed by the human eye,and presented in a visual manner in front of the human eye,so that the real vibration state of the equipment can be seen at a glance.In machine vision-based vibration monitoring of moving equipment,video motion amplification technology has unparalleled technical advantages and great application prospects.This paper investigates the potential application of video motion amplification technology for vibration monitoring of dynamic equipment.The research work is carried out to address the problems of blurred images and motion artifacts caused by large motions being amplified within the vibration monitoring screen during practical applications,as well as the problem of traditional methods relying on manual setting of time-domain bandpass filter parameters,and the effectiveness of the research methods in this paper is verified.The main contents of the paper are as follows:(1)Machine In terms of system hardware,a high frame rate video acquisition system is set up using high-speed industrial cameras,strobe-free lighting sources,PC computers and other devices;in terms of system software,video acquisition software and image processing algorithms are configured.Using the built machine vision-based dynamic equipment vibration monitoring system to achieve video data acquisition,video data processing,output video processing results and other functions.(2)Rresearch on the method of artifact removal based on acceleration filtering and amplitude threshold line selection.Based on the formation mechanism of image blurring and motion artifacts,an acceleration filtering algorithm is used to separate the linear large motions,and the remaining nonlinear large motions are separated by amplitude threshold line selection,and finally the amplification of only small vibrations is achieved.In the method validation part,the research method is verified using public video data and experimental video data,respectively.The validation results show that adding the large motion and tiny vibration selection process before the vibration amplification processing can reduce the interference caused by the large motion being amplified,and effectively improve the tiny vibration amplification effect of the dynamic equipment.(3)Research on the detection method of minute vibration parameters based on the time-shifted phase difference.The proposed method is based on the variation of the image phase signal in the time sequence,and based on the principle of phase correction to accurately estimate the small vibration frequency of the equipment within the input video,and the detection results are used as the input parameters of the band-pass filter to complete the small vibration squared without human intervention.Finally,the proposed method is validated by designing rotor unbalance and misalignment fault experiments based on the rotor test bench.The validation results show that the proposed method can accurately detect the failure frequency of the rotor.