Fast and High Precision Alignment Algorithm Based on Multi – Vector
Dingjie Xu, Pan Jiang, Ya Zhang, Shiwei Fan, Guochen Wang, Harbin Institute of Technology, China
Marine Strapdown Inertial Navigation Systems (SINS) inevitably experience disturbing motion, even if the carrier ship is moored. In order to improve the combat response capability of the ship, it is of great significance to study the fast and high precision alignment under the condition of shaking. At present, in the traditional alignment method, the double vector alignment method based on the gravity vector in the inertial system projection has strong environmental adaptability and can be aligned under the moving base condition. However, due to the use of only two vector information, the effective information utilization is low. So the alignment accuracy is not high.
In order to realize the fast and high precision alignment of the ship, it is possible to improve the precision of the coarse alignment by enhance the information utilization rate by using the fast and high precision multi-vector alignment technology, thus shortening the time of fine alignment.
Firstly, the two intermediate axis sets decompose the attitude quaternion into three separate quaternions representing the following: true inertial rotations; transport and Earth-rate rotations; initial angular alignment. The Earth-rate rotations and the true inertial rotations can be obtained from local geographic information and gyro measurement information, respectively. We select the velocity information as a non-collinear vector pair for the unknown quaternions. The problem of solving the fixed attitude matrix by vector is the famous Wahba problem, and the solution to the unknown matrix is transformed into the solution of the Wahba problem. With the alignment process, more and more information can be used, the accuracy of alignment is getting higher and higher.
The accuracy of the computation of the optimal quaternion when the angle of rotation is close to ? may be further improved as follows. When the angle of rotation is greater than ?/2 , the rotation can be expressed as a rotation through ? about one of the coordinate axes followed by a rotation about some new axis through an angle less than ?/2 . An initial rotation through ? about one of the coordinate axes is equivalent to changing the signs of two of the components of each reference vector.
It is found that the multi-vector alignment algorithm proposed in this paper is superior to the traditional double-vector alignment in alignment accuracy and alignment repeatability by simulating the static, level 2 sea, level 4 sea and level 6 sea condition. What’s more we also collected a number of practical shipborne data for off-line processing, based on multi-vector algorithm alignment results are all closer to the actual fine alignment results.
In summary, this paper presents a new fast alignment method which is effective to any movement form. Compared with the traditional Dynamic Base alignment method based on two vector, the new method can use the full information so we can use it to achieve higher alignment accuracy. We use the q-method which is used to solve the Wahba’s problem to achieve the attitude estimation. The experiment results show that this method can meet the quickness and satisfy the requirement of accuracy.