SIMULATION AND IMPLEMENTATION OF MANEUVERING TARGET TRACKING USING IMM-KALMAN FILTER ALGORITHM

• المؤلف الرئيسي: Yunita, Meutia
• التنسيق: Theses
• اللغة: Indonesia
• الوصول للمادة أونلاين: https://digilib.itb.ac.id/gdl/view/49613
• المؤسسة: Institut Teknologi Bandung

The tracking algorithm is used to predict the state (position, speed, acceleration) of a moving target that is within the range of a certain system, for example radar. The tracking algorithm must be able to estimate the state of the target with various types of movements, both constant and maneuverable. Until now, tracking algorithms on maneuverable targets is still an interesting topic to research. In this study, the developed tracking algorithm focuses on predicting targets in the form of aircraft and will be tested using real measurement data from the results of Automatic Dependent Surveillance – Broadcast (ADS-B) signal processing. Before testing real data, the algorithm will be tested on three types of trajectories generated in the simulation. These three trajectories describe the types of movement an aircraft may make including its maneuvers. The IMM algorithm is the most effective hybrid estimation system for predicting target maneuvers. This algorithm is widely applied to overcome various real cases for tracking maneuvering aircraft. In the IMM algorithm, the tracking process is carried out using several filter models in parallel using a soft switching or selfadaptation approach. In general, the IMM algorithm is developed using the target dynamic model of Constant Veocity (CV) and Constant Turn (CT) or CV and Constant Acceleration (CA). However, currently no one has discussed the comparison of the performance of these two algorithms for detecting target maneuvers. In this study, an evaluation of the two algorithms was evaluated in terms of their error values and their computation time compared to the single-mode Kalman Filter CV. In addition, the performance will also be compared with the IMM algorithm with CV, CA and CT target dynamic models. Based on the simulation and implementation results, it is concluded that the IMM-CVCT algorithm has the best error performance for all types of trajectories tested. From this research it can also be concluded that the IMM algorithm requires a much longer computation time than the single model Kalman Filter, but the increase in accuracy is much more significant than the increase in computation time.