An Improved Maneuvering Target Imm State Estimation Algorithm

Target tracking are widely applied in the field of national defense and civilian applications. With the improvement of flight vehicles' maneuverability, the target trajectories show a certain degree of complexity, randomness and diversity. Therefore, the study of maneuvering target tracking is a challengeable task for both the theory and application. The main purpose of designing tracking system is to track of the target reliably and accurately. Therefore, the study of maneuvering target tracking is of great theoretical significance and engineering application value.Presently, the researches of maneuvering target tracking are focused on interacting multiple model (IMM) estimator, and it has been proved to be the most cost-effective hybrid state estimation schemes. With theoretical analysis and experiment simulation, this paper shows that the proposed algorithm using constant velocity (CV), constant acceleration (CA) and the "current" statistical models is better than the classical IMM estimator using CV and CA models.In addition, the combination of theoretical analysis and the results of digital simulation show that the traditional IMM algorithm using CV, CA and the "current" statistical models also have some shortcomings, and then the corresponding improved algorithm is given.First of all, the model of IMM estimator will affect the overall performance of the IMM estimator. "Current" statistical model has a low accuracy for targets with weak maneuvering and fixed maximum-acceleration. IMM estimator using the "current" statistical model with variable maximum-acceleration can improve the overall performance of IMM estimator.Secondly, the transition probability of classical IMM estimator is determined by experience. In order to obtain more accurate posterior estimates and improve the accuracy of the model integration, it should make full use of the current information about the system mode and calculate the transition probability of Markov in real time.Finally, the classical IMM estimator can only deal with the flight path where the system noise and the measured noise are limited to be independent white noises with zero mean value. However, the data are often associated with colored noise instead of white noise in practical engineering. If the colored noises are deal with as white noise, the estimate accuracy will decline. Due to this reason, this paper considers the colored noise under the target tracking.After improving the IMM estimator from the three aspects as above, the improved IMM estimator is proposed, and Monte Carlo simulations are used to investigate several different flight paths. The digital simulation results show the feasibility of the proposed algorithms, and which could supply a reference for the application in engineering.