Monocular vision-based dynamic calibration method for determining the sensitivities of low-frequency tri-axial vibration sensors

Abstract

Low-frequency vibrations exist widely in the natural environment and in human activities. Low-frequency tri-axial vibration sensors are enormously applied in the fields of seismic monitoring, building structure health monitoring, aerospace navigating, etc. Their sensitivity calibration accuracy directly determines whether their applications can work reliably. Although the laser interferometry recommended by the International Standardization Organization (ISO) is commonly used to achieve the vibration calibration, it suffers from the shortages of low-frequency range, high cost, low efficiency, and limited applicable environment. In this study, a novel monocular vision-based dynamic calibration method is proposed, which determines the whole sensitivities of tri-axial sensors by the monocular vision method to accurately measure the spatial input excitation. This method improves the calibration performance by eliminating the installation error and enhancing calibration efficiency via decreasing reinstallations. The experimental results compared with the laser interferometry demonstrate that the investigated method can obtain similar calibration accuracy in the range of 0.16-2 Hz with more efficiency. The corresponding maximum relative deviations of X-, Y-, and Z-axial sensitivities were approximately 2.5%, 1.8%, and 0.4%. In addition, the maximum relative standard deviation of the investigated method was only about 0.3% in this range.