Integrated Navigation Approach Research Based On Fusion Algorithm

The navigation accuracy is low and fault-tolerant performance is inferior when we use asingle sensor in the navigation application, especially using the low-cost sensor. In thiscondition, we should employ multi-sensors to solve this problem, and fuse the measure datebased on some integration algorithm, to make the bias be constrained and advantages becomplemented by each other. This method can be named integrated navigation. For the aim ofgetting better navigation result, some researches were done as follows in this paper.1Analyzed the navigation principle of strapdown inertial navigation systems (SINS), andexplored the error distribution of SINS. Discussed the positioning principle of GPS on thebasis of pseudo range and pseudo range rate, and then analyzed the key technology of thecarrier phase measure method. And at last, the positioning and attitude estimating principle ofcelestial navigation and earth geomagnetism navigation systems were researched and theiranalyzed their error sources.2On the basis of least square filtering theory, analyzed the mathematical principle of kalmanfilter and its recurrence method, then discussed the advantages and disadvantages of extendedkalman filter, on this basis, unscented kalman filter course was discussed, too. After this, thesuperiority of some nonlinear principle such as adaptive filter and strong tracking filter theory,especially in the condition of non-Gaussian non-linear environment, were introduced. Then,presented a new filter algorithm, that is cubature filter, introduced its mathematical theory indetail. On the above, analyzed and compared the stability performance of cubature filter andunscented kalman filter, and compared the computing amount of cubature filter and unscentedkalman filter based on computer simulation. On the base of strong tracking principle,presented a novel strong tracking cubature filter. Then, researched particle filter, and raisednovel particle filter algorithm based on particle swarm optimization.3For the aim of fusing several navigation sensors’ output date, researched federated filter model which used widely. In the condition of multi-sensors measuring, analyzed the effect ofobservability for fusion result, and discussed the distribution factor and its action course,analyzed the rationality of vector distribution factor elementary. To improve the engineeringapplication performance, presented a novel cholesky decomposition method to accumulate theinverse matrix.4On the base of simulation, tested the fusion accurate of the novel federated filter whichraised in this paper and the result is higher than the traditional federated filter.5Explored neural network based GPS/INS integrated navigation method primary, and usedparticle swarm optimization in the training course. Then compared the navigation result ofneural network assistance and without neural network the result shows, neural network can beused to reduce the navigation error on some scale.6Reviewed the main work of this paper, and summed up the problem which haven’t beenexplored and dealt effectively. Meanwhile, looked into distance of application and developdirection of integrated navigation.