D.S. Feoktistov – Post-graduate Student, Siberian Federal University (Krasnoyarsk); Design Engineer, JSC «SPE «Radiosvyaz» (Krasnoyarsk)
The wide application of the Kalman filter algorithm in modern information systems is due to a number of its advantages. First of all, the current estimate formed by him is the best in the sense of the variance minimum in comparison with other linear estimates
obtained only by linear transformations of observations. In addition, several complicating the algorithm of the extended Kalman filter, it is possible to use it in problems of nonlinear filtering. The step-by-step (recurrent) nature of the Kalman algorithm makes it possible to obtain the current estimate by adjusting its previous value using only the next observation. This is useful for implementing a digital filter in real time, i.e. as data becomes available.
The aim of this work is the study of adaptive Kalman filtering of the various models of the motion of surface objects. The task set in the study is to develop an adaptive Kalman filter with adjustment of filtering parameters to the current mode of operation (motion model). On the one hand, the filter should not miss the maneuver, i.e. have sufficient sensitivity to changes in the trajectory, and on the other – as little as possible to respond to the disturbance in the radio navigation information when the object moves in a straight line.
When developing the Kalman filter for marine radio navigation systems, the object motion model plays a crucial role. In General, complete models of vessel motion are dynamic, representing the differential equations of motion of the center of mass of the vessel and the kinematic equations of communication of speeds with angular and spatial coordinates. To estimate the motion of an object on the plane, it is enough to use simpler models that would reflect the main changes in the parameters of the object movement. Changing the course or speed of a vessel can take, depending on the manoeuvrability of the vessel, from a few seconds for small vessels (boats) to several minutes for large vessels with high inertia (tankers). In a situation where you urgently need to change the course or speed of the ship, this delay can be critical.
Based on the simulation results, it can be concluded that the use of optimal Kalman filter estimation allows to reduce the coordinate skew by 1,5…3 times for the considered trajectory. The deterioration of the Kalman filter efficiency at site B is due to the sharp maneuvering of the object without reducing the speed of movement, which led to the distortion of the trajectory by 3 discrete counts, which can be explained by the insufficient speed of the filter reaction.
- Semushin I.V., Cyganova Yu.V., Kulikova M.V., Fat'yanova O.A., Kondrat'ev A.E. Adaptivnye sistemy fil'tracii, upravleniya i obnaruzheniya / Kollektiv. monografiya pod. red. prof. I.V. Semushina. Ul'yanovsk: Izd-vo UlGU. 2011.
- Spravochnik po teorii korablya: Gidrodinamika. Soprotivlenie dvizheniyu sudov. Sudovye dvizhiteli. V 3–h tomah / Pod red. Ya.I. Vojtkunskogo. L.: Sudostroenie. 1985. T. 1.
- Lukomskij Yu.A., Peshehonov V.G., Skorohodov D.A. Navigaciya i upravlenie dvizheniem sudov. SPb.: Elmor. 1996.