基于最优扩维法的现代目标跟踪系统非线性估计问题研究

Modern target tracking systems are faced with tremendous challenges arising from the new generation of military aircrafts and ships which have come into service. With the development of sensor technologies and the improvement of target capabilities, modern target tracking systems exhibit highly complicated measurement nonlinearities and motion nonlinearities. This project is focused on the hard but urgent nonlinear estimation problem in modern target tracking systems. To do this, this project first proposes an optimal augmentation framework, to generally formulate and solve the above nonlinear estimation problem. Within this framework, the nonlinear estimation problem for highly accurate estimates is researched considering the above two kinds of nonlinearities, respectively. The estimator based on the measurement augmented by its optimal uncorrelated conversion is proposed to solve the estimation problem caused by the measurement nonlinearities. And the variable-structure multiple-model estimator based on the optimal model augmentation is also proposed to solve the estimation problem caused by the motion nonlinearities. These two estimators can be integrated within the optimal augmentation framework to solve the estimation problems in which the two kinds of nonlinearities simultaneously exist. Based on this, the method of the optimal statistical linearization of augmented measurements is proposed, to solve the nonlinear estimation problem for extended object tracking. The proposed optimal augmentation framework and the related approaches have generality, optimality, and also the expansibility in improving the estimation performance continuously. The progress and breakthrough attempted to be acquired by this project can provide a new theoretical framework and novel technical methods for nonlinear estimation, which has important practical significance for the performance improvement of modern target tracking systems.

新一代军用航空器及舰船的服役使得现代目标跟踪系统面临严峻挑战。随着传感器技术的进步和目标性能的提高，现代目标跟踪系统呈现出高度复杂的量测非线性和运动非线性。本项目旨在研究现代目标跟踪系统非线性估计这一关键难题。首先提出最优扩维法，建立统一描述和解决该非线性估计问题的理论框架。并在该框架下，针对上述两类非线性条件下的高精度估计问题分别进行研究：针对量测非线性，提出基于最优不相关变换扩维法的非线性估计器；针对运动非线性，提出最优模型扩维法变结构多模型估计器。二者可在最优扩维法框架下有机结合以解决两类非线性并存时的估计问题。在此基础上，本项目提出扩维量测最优统计线性化方法，以解决扩展目标跟踪非线性估计问题。上述最优扩维法及相关方法具有通用性和最优性，以及可持续提升性能的可扩展性。本项目拟取得的进展和突破，可为非线性估计提供新的理论框架和技术手段，对于提升现代目标跟踪系统性能具有重要的实际意义。

随着新一代军用航空器及舰船的服役，现代目标跟踪系统面临严峻挑战。传感器技术的进步和目标性能的提高使得现代目标跟踪系统呈现出高度复杂的量测非线性和运动非线性。其核心难题即为非线性估计问题。因此，研究并提出一种具有通用性和可扩展性的非线性滤波理论及算法，不仅可对估计理论进行有益的扩充，也具有重要的应用价值。本项目针对现代目标跟踪系统非线性估计问题进行研究，取得了如下进展：. 1）针对非线性估计问题，建立了最优扩维法理论框架，既可统一描述现有流行的非线性估计器，也可通过扩维提高估计精度或简化非线性估计问题，为后续研究提供了理论基础；. 2）在最优扩维法框架下，提出了最优不相关变换扩维法，并在此基础上提出了仅需前两阶矩的点估计器，突破了传统的线性最小均方误差估计器的线性结构制约，估计性能进一步提高；. 3）基于最优扩维法框架，提出了实时、综合而兼具最优性的目标状态及形态联合估计机制，并在贝叶斯框架下推导出相应的机动扩展目标估计算法，较好解决了超/高机动扩展目标的跟踪问题；. 4）基于最优不相关变换及机动扩展目标估计方法，在随机矩阵框架下，对机动目标的运动模式和量测进行建模，提出了扩展目标非线性状态及形态估计算法，解决了高机动扩展目标跟踪非线性状态及形态联合估计问题；. 5）上述算法提供的信息可在目标分类、捕获及故障诊断等方面应用。本项目提出了相应的联合跟踪与分类、捕获及故障诊断等方法，充分利用了估计与决策间的高耦合性，一体化提升性能；. 6）估计性能评估及排序研究对正确评价与提升非线性估计算法的性能有重要意义，本项目提出了一系列方法，能够给出算法的正确评估结果及不同算法间的合理排序。. 本项目取得的进展和突破，为非线性估计提供了新的理论框架和技术手段，对于提升现代目标跟踪系统性能具有重要意义。所提出的方法可通用于解决超视距目标跟踪及水下目标跟踪等实际应用中的非线性估计问题。