Dynamic parameter identification of a high g accelerometer based on BP-PSO algorithm

Dynamic calibration is an important measurement method to improve the dynamic performance of sensors. However, the structure and packaging of the accelerometer and the ubiquitous noise interfered with the results. Therefore, a method based on BP neural network and particle swarm optimization algorithm is proposed to dynamically calibrate the elevation accelerometer in a composite package in the frequency domain, to realize the systematic identification of the accelerometer dynamic model. Numerical simulations and experimental results demonstrate that compared with the narrow pulse width calibration theory, this method achieves the accuracy requirement while broadening the calibration band by more than 7 times. It is also proved that this method is effective in parameter identification.