Marine Radar Target Tracking Algorithm And Performance Analysis Based On Kalman Filter

In the past, target information was displayed on the interface of marine radars. In order to understand the target moving state, the information must be manually plotted and calculated to determine the target's moving direction and speed. With the extensive application of microprocessor, marine radar systems can quickly complete filtering, tracking, prediction and so on. So it can provide collision avoidance warnings to the navigation officers. The tracking filter is an important component of the system. With the mobility of the goal becoming stronger, the requirements of filter's accuracy and stability are getting higher and higher. The main task of this thesis is to design the adaptive filters which have high accuracy, real-time tracking, and the ability to adapt to target maneuver quickly.In the first place, this thesis introduces some basic filtering algorithms such as Kalman filter, a-P filter, andα-β-γfilter and so on. At the same time the thesis analyzes their characteristics and nature of the algorithm theoretically. Then a basic Kalman filter is designed for marine radar target tracking, the effect of the disturbance noise vector and observation noise vector of white Gaussian noise is discussed. Because that the parameters of disturbance noise vector and observation noise vector are unknown, and mobility of an object has the characteristics of random unpredictable, we must improve the filter in order to prevent filter divergence. There are many adaptive Kalman filter algorithms. In this thesis, velocity component is used to identify the disturbance noise variance of targets by method of statistical analysis. The distance component is also used to identify the observation noise variance of targets by method of statistical analysis. This method avoids the calculation of determining when the target maneuvering occurs. The calculation is simple and more suitable for real-time tracking. The performance of the adaptive filter is tested by simulation experiments.Theα-βfilter is the filter for uniform motion, and theα-β-γfilter is the filter for uniformly accelerated motion. Finally, this thesis simulates three filters in different states, and compares their performances, and discusses the differences of performances between Kalman filter and other filters.