Research on moving base alignment of ship-borne strapdown system based on wavelet and AI technology

This paper introduced a method of initial alignment of ship-borne SINS based on wavelet and ANFIS. It provided position, velocity and heading angle corrections for INS to prevent the error growing with time so that it improved the accuracy and robustness of the whole system. To achieve the object, method of wavelet multi-resolution analysis was used to de-noise the output of the sensors by comparing the wavelet coefficients of same resolution level and reconstructing high accuracy error signal. Positions or velocities and heading angle were set to be the input of ANFIS while the corresponding corrections were output after trained. The desired output of trained samples was position, velocity and heading angle error which is the result of the wave multi-resolution analysis. So wavelet multi-resolution analysis and ANFIS complement each other in this paper. The result of simulation test shows that the novel method can speed up convergence and improve accuracy. The error of horizontal alignment is within 1´ and that of azimuth alignment is within 3´ while the trend of accuracy using kalman is divergent. It is shown that this method can provide reliable and effective alignment for ship-borne SINS.