Research On The Key Technologies Of Machine Vision-based Low-frequency Vibration Calibration

With the continuous improvement of vibration measurement accuracy,the demand for high-accuracy vibration calibration is increasingly prominent.The vibration sensor provides vibration measurement data for on-line monitoring,optimal control and decision,etc.However,affected by the mechanical manufacturing and the sensor material aging,the actual sensitivity of the sensor is different from its nominal sensitivity,which reduces the accuracy and reliability of the measurement data.Therefore,it is of great significance and engineering value for earthquake warning,bridge and building structure testing,wind power safety monitoring,mechanical fault diagnosis to develop the high-accuracy vibration calibration method which ensures the effectiveness of the measurement data.The dissertation is on the basis of detailed analyzing the vibration calibration principle and the machine vision vibration measurement methods.Aiming at the problems of current calibration methods on the limited measurement accuracy of low-frequency vibration excitation acceleration,the low sensitivity calibration accuracy of low-frequency vibration sensor,the limited calibration efficiency of multi-axial vibration sensor,and the complex calibration system,the researches are mainly conducted as follows:1.A machine vision-based single-component linear vibration calibration method was proposed.The reliable camera calibration method,the blurred edge enhancement method for motion sequence images,and the sub-pixel edge extraction method were adopted to improve the excitation acceleration measurement accuracy of the low-frequency vibration sensor,and then achieves the high-accuracy sensitivity magnitude calibration.Meanwhile,the machine vision method was used to measure the bending of the excitation generator's guideway and correct the bending in order to further improve the sensitivity magnitude calibration accuracy.The time-spatial synchronization-based sensitivity phase calibration method was proposed.The output signal of the sensor and its excitation acceleration signal measured by the machine vision method were spatially aligned by utilizing the zero-encoder to accurately calibrate the sensitivity phase.The proposed calibration method can accomplish the high-accuracy sensitivity magnitude and phase calibra-tion in a broad low-frequency range,especially for the improvement of the calibration accuracy at lower frequencies.2.A machine vision-based two-component linear vibration calibration method was proposed.The machine vision method was used to decoupling measure the excitation displacement provided by the two-component shaker to get the excitation acceleration and orbit of the multi-axial vibration sensor.A plane sensitivity model was built to accurately describe the sensitivity of this sensor,and an elliptic orbit excitation acceleration-based plane sensitivity calibration method was proposed.The axial sensitivity magnitude and phase,transverse sensitivity of the multi-axial sensor were calibrated by combining the time-spatial synchronization technology.Experimental results show the proposed method can efficiently and accurately calibrate the plane sensitivity of the multi-axial sensor.3.A virtual traceablity method for the machine vision linear vibration calibration was proposed.The standard sinusoidal vibration excitation generated by the computer was used to reduce measurement uncertainty sources of the machine vision method.The quantity-value traceability system of the machine vision vibration calibration method was built by evaluating the uncertainties of these sources.This method shortens the traceability error chains of the machine vision method and realizes the quantity-value flat transfer.4.An improved Earth's graviation method was presented.The machine vision method was adopted to measure the Earth's graviation field direction and rotation angle in order to improve the excitation acceleration measure-ment accuracy.The high-accuracy sensitivity magnitude and phase calibration of the Earth's graviation method was achieved by applying the time-spatial synchronization technique.Aiming at the problem of heterodyne interfero-metry has to collect large amounts of data at low-frequency.The bandpass sampling rate was adopted to collect the heterodyne interferometer signal,and the optimal sampling rate calculation method used for the high-accuracy low-frequency vibration calibration was provided.In the low-frequency range,the proposed machine vision-based single-component,two-component linear vibration calibration methods,the laser interferometry,and the Earth's graviation method were used to calibrate the vibration sensors.Compared with the Earth's graviation method,the machine vision method improves the sensitivity phase calibration accuracy and extends the upper limitation frequency of calibration frequency range.Compared with the laser interferometry,the machine vision method improves the sensitivity magnitude and phase calibration accuracy in the lower frequency,decreases the complexity of the multi-axial vibration sensor calibration system and the uncertainty caused by the repeated installation.The machine vision calibration method has the advantages of low-cost,flexibility,high-accuracy calibration in a broad low-frequency range,etc,which can be used for field vibration calibration and has important engineering application value.