Research On Integrated Navigation Technology Of Strap-down Inertial Navigation And Ultra Short Baseline

As people rely on the underwater vehicles to probe into the ocean,how to achieve highprecision navigation has become a hot research topic.SINS can provide complete navigation information for the autonomous navigation of the vehicle,whose error,however,accumulates with time,making it impossible to meet the needs of long-term navigation;as a simplified acoustic positioning device,USBL can be installed on the underwater vehicles to provide position information for it,but its working range is limited.With the method of inertial integrated navigation,the accumulated error can be corrected by USBL,to improve navigation accuracy.Thus,the methods to improve the accuracy and stability of SINS and USBL integrated navigation are studied in this paper.Aiming at implementing SINS/USBL integration,the loose and tight integration based on extended Kalman filter are designed.The position output of USBL is applied in loose integration,while the TOA and TDOA of USBL are employed in tight integration,to correct inertial navigation output by indirect filter.The simulation results show that the accumulated error of SINS can effectively be restrained with the two integration methods,along with low error convergence of the celestial position and velocity in initial phase of tight integration and low accuracy of loose integration;after the filtering is stable,the navigation error of loose integration still increases when the array opening angle is close to 90°,while that of tight integration is stable,with the overall accuracy more than 15% higher than that of loose integration.Aiming at addressing the time-varying problem of the observation noise,the Sage-Husa adaptive filtering algorithm based on sequential filtering is studied.Determining the dependence of the current noise on the past noise with a weighting factor,the algorithm estimates the covariance matrix of the observation noise combined with the innovation.The simulation results show that Sage-Husa algorithm,after used in loose integration,has fast convergence speed and integer of the navigation error,but the navigation error changes little after used in tight integration.Aiming at the problem that the error of loose integration is greatly affected by the USBL opening angle,a method is proposed to control the filter gain with the cosine of the opening angle as the weighting factor when the opening angle is greater than 80°.Additionally,the simulation results show that the accuracy of the method is 19% and 45% higher than that of Sage-Husa algorithm,respectively.According to the measurement characteristics of USBL,this paper proposes to introduce the radial velocity measured by USBL into the loose and tight integration.The simulation results show that the introduction of radial velocity can contribute to improving the horizontal navigation accuracy of loose and tight integration,while the effective of the celestial navigation accuracy is poor.Therefore,to solve this problem,a combined mode switching method is proposed,That is,the data fusion is carried out with the horizontal position in the array coordinate system and radial velocity when the opening angle is greater than 80° and the USBL positioning is invalid,along with the use of the loose integration of radial velocity in other cases.The simulation results show that the celestial velocity and position accuracy of this proposed method is 65% and 58% higher than that of the loose integration of radial velocity,and 66% and 25% higher than that of the tight integration of radial velocity,respectively.