Research On The Models And Methods Of Ultrawideband/gnss/sins Integrated Positioning

The combination of Global Navigation Satellite System(GNSS)and Strapdown Inertial Navigation System(SINS)can provide continuous and reliable navigation and positioning services in outdoor environments and/or in the situation when satellite signals are temporarily outage.However,for indoor spaces or indoor-outdoor transition areas,a reliable positioning services will be difficult to maintain owning to the blockage or severely missing of satellite signals for a long time.Ultra-wideband(UWB)system can provide centimeter-level ranging accuracy in theory and has the advantage of simple deployment.It can provide effective ranging information to satisfy positioning services for indoor and/or outdoor areas where GNSS signals are weak.Based on land vehicle navigation,this dissertation focuses on models and methods of indoor and outdoor positioning,covering the researches on models and methods of the integration of UWB/SINS,GNSS/SINS,and UWB/GNSS/SINS.The key researches of the dissertation are summarized as follows:(1)The UWB base station coordinates are usually determined by measurement in advance,which inevitably has a certain deviation from the theoretical true(but unknown)coordinates.Therefore,for the integration of UWB/SINS,a dynamic EIV(Errors-in-Variables)model is proposed,which takes the UWB base station coordinate error into account.A total Kalman filter method is proposed based on the dynamic EIV model.The theoretical deviation formula of the a posteriori state is derived and the effectiveness of total Kalman filter method on dynamic EIV model is analyzed based on the theoretical deviation formula.The results show that,in practical applications,considering the error of UWB base station coordinates does not necessarily improve the positioning accuracy of UWB/SINS.The positioning results are affected by the comprehensive influence of UWB base station layout,base station accuracy and the grade of SINS.Furthermore,the indoor unmanned vehicle experiment verifies the effectiveness of the proposed method.(2)In view of the fact that the state-space model of UWB/SINS is usually nonlinear,the expression of the nonlinear dynamic EIV model is given.A generalized total Kalman filter method for the nonlinear dynamic EIV model is derived by using Gauss-Newton method and the computational complexity of it is also analyzed.The results show that the generalized total Kalman filter method can handle the nonlinear dynamic EIV model,and its computational complexity is slightly higher than that of the extended Kalman filter method.(3)On the basis of the nonlinear dynamic EIV model,an unscented total Kalman filter method is derived.Since this method needs to generate a large number of sampling points(sigma points),the nonlinear transformation of each sigma point increases the computational cost.The following two strategies are adopted to increase the computational efficiency: 1)conditional linear transformation + marginalized unscented transformation: transforming the original state space model into a conditional linear model where the transformation is linear in states and nonlinear in others,and then the sigma points are generated only for the nonlinear-related states,reducing the sigma number of sigma points;2)parallel operation: sigma points are allocated to multiple CPU cores at the same time to perform the transformation of sigma points in parallel aiming to achieve the real-time calculation.(4)For GNSS/SINS integration in outdoor environment,owing to road conditions and an inertial measurement unit(IMU)on the carrier not being rigidly mounted,biased measurements in the IMU will occasionally emerge,leading to discontinuous and unreliable ambiguity fixing results.It is proposed to introduce positional polynomial fitting(PPF)constraint to the GNSS/SINS to aid ambiguity fixing.The PPF predicts the position of next epoch based on the position information in time window as well as a predefined model order,and its prediction is independent of the abnormal kinematics between epochs.It is proposed to use model probability and ambiguity fixing state to judge whether an IMU has biased outputs comprehensively.If it exists and affects the normal ambiguity fixing,the PPF is triggered to aid the ambiguity fixing and update GNSS/SINS tightly coupled solutions.The results show that the PPF-aided method can effectively compensate for the epochs where correctly fixed ambiguities is impossible due to biased outputs of an IMU;it is also found that the partial ambiguity fixing strategy is helped to achieve more reliable results in combination with the PPF-aided dual-constellation and INS integration.(5)Integrating UWB ranging information will help improve the positioning reliability and accuracy of GNSS/SINS.Taking UWB ranging information as equality constraints and mining the internal/external hidden information of GNSS/SINS according to actual application,a UWB/GNSS/SINS integrated model with equality and inequality constraints can be established.With inequality constraints,the state estimates are usually obtained by means of a search method where one can only obtain a non-explicit estimate(no analytical solution),therefore the computational efficiency will be low and the precision estimation of estimated state cannot be evaluated.It is proposed to apply the aggregate function method to the inequality constrained Kalman filter.The aggregate function method can transform all the inequality constraint equations into a single and smooth nonlinear equality constraint equation.Hence,the Lagrangian multiplier method can be directly used to calculate the state and precision estimation(with analytical solution),which avoids a time-consuming search method and is more suitable for real-time navigation calculation.The results show that the proposed method can achieve similar results with the search method(take the sequential quadratic programming(SQP)as an example),however,the time cost is reduced by nearly 10 times compared with the SQP method.The dissertation contains 44 figures,13 tables,and 221 references.