Damping And Recalibration Of Inertial Navigation System In USV Collaborative Network

In collaboration of fleet,the main boat is equipped with high precision inertial navigation system,in order to save costs,the slave boat is equipped with low precision MEMS navigation system.The main boat’s high precision navigation information can be shared through the damping,comprehensive correction,nonlinear filtering and collaborative algorithm,so as to improve the positioning accuracy of the slave boat.The paper mainly discusses the constant velocity feedback,lead,lead lag and two order matching horizontal velocity damping network design methods,and analyzes the effect of external velocity error,gyrodrift and accelerometer bias on the damped inertial navigation system,then the horizontal velocity damping network is added in the strapdown inertial navigation system formula.Furthermore,the design of damping network based on external positioning error and horizontal attitude network are also discussed.When the speed error is large,the inertial navigation system in the slave boat should be switched to undamped state by means of kalman filter.Compared with horizontal damping networks,damping-ratio-adjus table azimuth damping network is designed.Network transfer function is only related to the damping ratio.The design method is more regular in order to avoid the trouble of traditional design method according to Potter diagram.The two-point correction method can be used when the slave boat can simultaneously obtain the position and heading.Three point calibration can be used if only the location information is available,which can calculate the drift of MEMS gyro in three axis of slave boat.The random calibration is carried out in slave boat according to the Ψ angle error model and angle error model of inertial navigation,they can provide real time attitude error and gyrodrift.Due to the nonlinear characteristics of slave boat’s dead-reckoning formula,the MEMS gyro’s error can be corrected by different methods of nonlinear filtering.Decentralized augme nted information filter is suitable for distributed implementation of multi platform algorithm,in order to avoid computation burden is too concentrated.The extended Kalman filter(EKF)algorithm is simple,good real-time performance,and it has been widely applied in engineering practice,but Jacobian matrix’s solving inevitably leads to one-order approximate error.In contrast,unscented Kalman filter(UKF)and cubature Kalman filter(CKF)can be accurate to Taylor expansion’s three order or higher order terms of nonlinear model,so the estimation accuracy and speed are better than EKF.Non-additive-noise filter is more suitable for the noise characteristics in the navigation equation.Collaborative algorithm can estimated slave boat’s position in the absence of the heading information conditions,and its location information can be used as measurement in integrated navigation with inertial navigation system.Because its measurement noise variance is needed,so error propagation characteristics of collaborative algorithm must be deduced.Because of the large inertia and poor mobility of slave boat,curve fitting can be used to estimate its current position by its past locations,this method has good effect when slave boat sails uniformly and directly.