Novel GPS/SINS Integration Architechture And Systematic Error Compensation Methods

For the conventional GPS/SINS integration techniques, there exists a basic dilemma, namely, the high precision integrated systems generally have poor reliability and real-time performances, and on the other hand, the systems with good reliability and real-time performances generally have low precision. For example, the integration of GPS RTK/SINS can be very precise, however, its reliability will become poor with the increasing of the distance from the GPS base station and its real-time performances are not good because of the necessity for the resolution of the integer ambiguities. So, generally, GPS RTK/SINS integrated system can not be used in real-time or wide-area navigation applications. As an opposite example, the integration of GPS pseudorange/SINS is suitable for navigation applications because of its good reliability and real-time performances, however, its precision is low. So, the first goal of this research is to improve the overall performances of GPS/SINS integrated system by rearrangement of the GPS information and by using new integration techniques. Besides, the stochastic errors of the inertial sensors and the initial alignment error of the SINS are two kinds of the most crucial errors, so the second goal of this research is to develop some techniques to optimally compensate both of them with the aiding of GPS. To accomplish these two goals, the following work has been finished in this research:1. A new GPS observable, incremental carrier-phase observable, has been derived based on GPS carrier-phase, and the corresponding observation equation for the incremental carrier-phase has also been derived. Furthermore, the applications of the incremental carrier-phase in navigation and integrated navigation have been discussed. Error analysis has shown that the incremental carrier-phase is an unambiguous and precise observable with the advantages of both pseudorange and carrier-phase. Navigation experiments have revealed that the mm/s level accuracy has been achieved in computing velocity from the incremental carrier-phase. Integrated navigation experiments have demonstrated that the integration of the incremental carrier-phase and SINS can obtain very high short-term navigation accuracy, while the position error will slowly accumulate with time.2. For the optimal estimation systems with two kinds of observables whose measurement noises are quite different with each other, a novel optimal estimation technique, dual-rate Kalman Filtering technique, has been proposed, and the relevant theoretical analysis has been done. The dual-rate Kalman Filter fuses the low-noise observable with high rate, and the high-noise observable with low rate into the optimal estimation systems. By analyzing the propagation of the error covariance, it has been found that the proposed dual-rate Kalman Filtering can effectively isolate the noises from different error sources, and at the same time, the dual-rate Kalman Filtering can be used to fuse the asynchronous observations from different sensors and can also improve the computation efficiency.3. By using the dual-rate Kalman Filter as the integration tool, a novel tightly coupled integration system based on the incremental carrier-phase, the pseudorang, and the SINS has been designed. To implement the above integrated system, the system equation for the dual-rate Kalman Filtering has been derived based on the SINS error equation, and the measurement equation for the high-rate Kalman Filtering as well as the measurement equation for the low-rate Kalman Filtering has also been derived based on the incremental carrier-phase and the pseudorange, respectively. Experimental tests have demonstrated that, compared with the conventional integration techniques, the proposed novel integration technique has improved the position accuracy, and dramatically improved the velocity and attitude accuracy as well as the coasting performance.4. Colored noises have important influences to the performances of inertial sensors, and to compenstate the errors of the inertial sensors introduced by the colored noises in GPS/SINS integration, it is necessary to model the colored noises. Firstly, for a single colored noise, a general procedure has been proposed to derive its stochastic differential equation model from its power spectral density (PSD) function, and the stochastic differential equations of several common colored noises have been derived. Then, for multiple concurrently existing colored noises, an equivalence theorem relating to the stochastic differential equation has been proposed and proved. The equivalence theorem states that multiple stochastic differential equations can be equivalently expressed as a single stochastic differential equation in the wide sense stationarity, and gives the formulas to determine the coefficients of the equivalent stochastic differential equation. The equivalence theorem is the basis of the equivalent modeling and on-line compensation for multiple concurrently existing colored noises.5. An on-line compensation scheme for the stochastic errors of the inertial sensors has been proposed with the aiding of GPS. In the compensation scheme, the white noise, the quantization noise and the colored noises are dealt with as follows: (1) the white noise is used to determine the covariance matrix of the process noise; (2) by modifiying the SINS error equation, the quantization noise is converted into white noise, and is used to augment the covariance matrix of the process noise; (3) the equivalent stochastic differential equation model for multiple colored noises, which is derived from the equivalence theorem, is used to augment the system equation of the integrated system. The results of the experimental tests have demonstrated that the proposed scheme has effectively compensated the stochastic errors of the inertial sensors, and as a result, dramatically improved the performances of the pure SINS.6. To reduce the in-motion alignment error of SINS introduced by the modeling error under the condition of large initial heading error, a novel in-motion alignment approach has been proposed for the SINS. To implement the above approach, the system equation and the measurement equation for the coarse alignment and fine alignment have been derive respectively. Experimental tests have shown the superiority of the proposed alignment approach, and compared with the conventional approaches, the proposed approach has dramatically improved the accuracy and speed of the alignment procedure.7. To reduce the in-motion alignment error of SINS introduced by the observable error, a novel in-motion alignment scheme has been designed with the aiding of high precision incremental carrier-phase. Experimental tests have indicated that the alignment accuracy of the SINS with the aiding of the incremental carrier-phase has almost approached the alignment accuracy with the aiding of GPS PPK (Post-Processing Kinematic) positioning.