Research Of A GPS/SINS Ultra-Tight Integrated Navigation System Based On Time-Space Difference

As one development direction of navigation system,the integrated navigation system including satellite navigation systems and inertial navigation systems is under increasing attention in the field of aviation and navigation.With the development of technology and high performance carrier required by the navigation system that has more accurate precision,more inexpensive cost,more trustworthy reliability.To satisfy these requirements the GPS/INS ultra-tight integrated navigation has gradually emerged as an important research topic.At present,a variety of GPS/INS ultra-tight integrated navigation systems models have been proposed,where the ultra-tightly integrated navigation system based on vector-tracking model is better than others.However,this model also has many problems,such as the model error,residual time errors,heavy computational load caused by the nonlinear model.To solve the problems above,an intensive study about GPS/INS ultra-tight integrated navigation systems is performed,and the main works are as follows:To solve the problems of the existing ultra-tight integrated navigation system model,a novel model named GPS/SINS ultra-tightly integrated system model based on time-space difference is presented.The new model eliminats the time delay and clock error by using the difference method.Moreover,the new model is built by using I/Q angular rate and phase information which is a linear function of the carrier’s velocity and position.Therefore,there are no linearizations or other approximate treatments,which leads to a small model error and less complicated model.A single observability analysis method cannot achieve a comprehensive and accurate relult.Therefore,a multiple observability analysis method is used in this thesis,which analyzes the observability of the system states in both sides and gets a qualitative and quantitative analysis result.Then,according to the analyzed results an overall presentation of system states is obtained.Finally,the proposed model in this work is as an example to verify the multiple observability analysis method.To alleviate the computation load,a dimensionlity reduction model of ultra-tight integrated navigation system is proposed.And according to the time-variation characteristic of the system noise,a corresponding improve interacting multiple model algorithm is designed.As the lever arm error is not a constant value,it is potentially time-variant,in this thesis we use the white noise to describe it.Therefore the lever arm error can be removed from the state,and the dimensionlity reducted model can be obtained.We use the interacting multiple model(IMM)algorithm to solve the problem of noise time-varying.To track the time-varying noiseit fuses the estimation of each model with different intensity noise.Taking into account the shortcomings in the traditional IMM algorithm that the sojourn time of each model must be known to determine the mode transfer matrix,we proposed the optimal mode transition matrix interacting multiple model algorithm(OMTM-IMM).And the parameters are divided into 6 different situations,and 6 improved IMM algorithms are obtained.Finally,theoretical derivation proves that the improved algorithms has higher estimation accuracy,besides,the simulation results confirm this conclusion.The feedback information of traditional revising filter method is blind which will lead to a bad influence to the navigation accuracy.To solve this problem,a revising filter method based on the observability is proposed.The new method computes the feedback factor by using the observability data mentioned above which means the feedback information can be adjusted according to the observability of system state.Therefore the problem of blind feedback can be solved and the increasing or divergence of the estimation errors can be avoided.Simulation results demonstrate that the improved method can reduce the estimate error and variance,thus improve the estimate accuracy of system.