Research On Maneuvering Target Tracking Algorithms Based On Multiple Models

Maneuvering target tracking is a challenging problem in the field of target tracking, and it has wide application prospect in civil and military areas. Thus the research and development of the maneuvering target tracking has received widely attention, and many researchers at home and abroad has done deep research on it and got rich results in recent years. Due to the diversity of maneuvering target movement ways, so the movement process of maneuver target cannot be well depicted by the single model filters, e.g. kalman filter, particle filter and so on.For this, the multiple model algorithms were proposed to solve the deficiency of single model filter algorithms, and significantly improve the tracking accuracy. Interacting multiple model as a classic algorithm of multiple model algorithms, because of the high cost effectiveness and good tracking performance, it has been widely applied to many practical scenarios.Therefore, we will study the interacting multiple model algorithm which is used for maneuvering target tracking in this article. Some improved algorithms are proposed, which can adapt to various situations of maneuvering target tracking, such as the sensor can receive deviation of the data, the clutter interference exists in target environment, and the nonlinear system. The work of this paper is divided into four parts, as follows:1. In order to solve the maneuvering target tracking in case of the possible bias of sensor,a novel IMM algorithm(IMM-ST-EV) is proposed in this paper. The proposed algorithm takes into account both the diversity of sensor measurement models and target motion models,by which can deal with the bias of sensor and the situation of target maneuver. However,considering the above two models together may result in too many models, thus degenerating the performance of tracking. To alleviate this problem to some extent, the extended Viterbi algorithm(EV) is incorporated into the proposed algorithm. Finally, the simulation results verify the effectiveness of this algorithm; meanwhile, the motion model matching problem owing to the possible bias of sensor and the maneuverability of target is solved by using multi-model characteristics.2. The PDA algorithm is incorporated into SIMM to become the SIMMPDA algorithm in order to improve tracking precision of maneuvering target in clutter environment. The PDA algorithm handles data association and measurement uncertainties in clutter. The SIMM deals with the model switching and receives the optimal state estimations of target in the linear minimum variance sense. In consideration of the problem that the matched model hasn't dominance obvious due to the interference of clutter at each time, so that the model probability of the SIMMPDA algorithm is modified. Thus, a M-SIMMPDA algorithm, which is a SIMMPDA algorithm based on model probability modification is presented. The simulation results show that the tracking accuracy of the proposed algorithm has been improved to some extent.3. In order to solve tracking problem of maneuvering target in nonlinear background, an Augmented Interacting Multiple Model Cubature Kalman Filter(AIMMCKF) algorithm is put forward here. To obtain the fixed-lag smoothing state estimation, IMMCKF approach is applied to a nonlinear state-augmented system in the proposed algorithm. At the same time, totackle different models problem within being represented in different state spaces,corresponding augmented conversion operation can be used, so that the model condition estimation can be in a common space for mixing and fusion. Therefore, the algorithm can be carried out normally. The simulation results show that, the proposed algorithm achieves higher precision and stronger adaptability for maneuvering target tracking in comparison with traditional nonlinear tracking algorithms.4. In order to effectively improve the performance of multi-sensors tracking maneuvering target under nonlinear condition, the SIMMCIF and DIMMCIF are proposed in this paper.These new algorithms are derived by incorporating the CIF algorithm into SIMM and DIMM algorithm separately, and a more general global information fusion method is used to deal with the multiple sensors information fusion problem. The SIMM and DIMM algorithms are used to deal with the mobility of target motion, and obtain the optimal state estimation of target in the linear minimum variance sense. The CIF use a set of cubature points of equal weight to solve the integration problem of Bayesian filter, thus linearization and calculating the Jacobi matrices are unnecessary. Moreover, CIF has simper update step. Simulation results show that the proposed algorithms have better fusion effect for the data which is received by asynchronous or synchronous multi-sensor, and the tracking accuracy has been improved to a certain extent.