Based On Nonlinear Filtering Of The Gravity Aided Navigation

The ocean is an important resource for human. With the development of technology,utilization of the oceans is getting bigger. In addition to using the Marine resources, it also plays an important role in military. People use marine,who inevitably make use of marine exploration tool,so the underwater vehicle came into being. With the development of computer technology and navigation technology, the navigation accuracy requirements of AUV is increasingly high. Underwater navigation, especially the military, in order not to let the enemy found, Concealment is crucial. In the face of this requirement ,countries made a lot of effort in this regard. From the celestial navigation to sonar navigation, but these still can't meet the requirements, until the terrain aided navigation, geomagnetic aided navigation,gravity aided navigation and various kinds of integrated navigation. In the navigation,gravity aided navigation is of great significance in underwater navigation,which can overcome the characteristics of the bad concealment. gravity aided navigation is faced with many problems,Gravimeter precision problem, the resolution of gravity digital maps. These problems made a great breakthrough in foreign countries, but our country still need to make a lot of research in this area. This paper was studied according to the theory of gravity aided navigation algorithm, meanwhile based on gravity aided navigation, which made deep research for nonlinear filtering algorithm. In this paper, the main work was done as follows:First:The main error of inertial navigation equations are deduced. Including position error equation, velocity error equation, the misalignment Angle error equation. According to the error equation, the state of gravity aided navigation model is established.Second: the extended kalman filter for weakly nonlinear systems and unscented Kalman Filter for strongly nonlinear system ,for the two process, the article has carried on the detailed instructions. on the basis of unscented Kalman Filter, Put forward a kind of extended unscented Kalman Filter.Third: according to the bayesian filtering algorithm, introduces a filter method for strongly nonlinear systems, FPKE filter method.Fourth: In this paper, the particle filter of the nonlinear filtering described in detail.From the traditional particle filter,monte carlo integration,re-sampling and so on,the advantages and disadvantages of particle filter are described.On the basis of the traditional particle filter ,introduced three kinds of improved particle filter. Two of these were improved according to the recommended density function.EKF and UKF were introduced into the particle sampling. On the basis of CKF,another filter were put forward in EM model to replace the gaussian mixture model.,which is called GMUPF particle filter algorithm.Fifth: In this paper, based on the classic terrain auxiliary method,because the terrain aided method and gravity method has a lot of similarities, our can combine classical method and EKF method, parallel computing, set the appropriate permissions,in every moment use appropriate auxiliary method. Another method is that ICCP method combined with filtering method. The ICCP method need more accurate navigation parameters, after filtering, the navigation parameter meets the request. The two methods of inertial navigation played a certain effect.