面向非圆信号的多维参数联合无源定位技术研究

Joint localization based on multi-dimensional parameters can extract more abundant information about the locations, which helps to provide the precise locations of targets for industrial production and military applications. Appling the unconjugated covariance matrix of non-circular signals in localization can further improve the localization performance. However, the research on passive localization for non-circular sources is not comprehensive, and thus their performance is limited. Therefore, this project investigates both the two-step localization and direct position determination (DPD) for non-circular sources, and mainly focuses on the localization techniques using multi-dimensional parameters: (1) In regard to two-step localization for non-circular sources, the joint estimation of multi-dimensional parameters based on extended covariance matching estimation technique and the performance analysis are studied. On this basis, the idea of the least squares localization based on calculation of the optimal weighting matrix is proposed. (2) With regard to direct position determination for non-circular sources, the joint direct position determination algorithm using multi-dimensional parameters based on weighted extended subspace data fusion and the error calibration method based on neutral computation are designed, which guarantee both the performance and computation ability. (3) The performance evaluation of direct position determination for non-circular sources and its quantitative analysis method are investigated. This project aims to enrich the theories and methods of passive localization for non-circular sources using multi-dimensional parameters, and thus provides new theoretical basis and technical support for improving the current technical level of localization for non-circular sources.

联合多维参数的定位方法能够获取丰富的定位信息，有助于为工业生产和军事应用提供准确的目标位置，在定位中利用非圆信号椭圆协方差不为零的特性，能够进一步提升定位性能。然而，目前关于非圆信号目标的无源定位研究不够全面，导致其定位性能受限。为此，项目从非圆信号的两步定位与直接定位两种方式展开研究，重点关注联合多维参数的定位技术：（1）针对非圆信号的两步定位问题，研究基于扩展协方差匹配估计技术的多维参数联合估计算法与性能分析，并在此基础上提出基于最优加权计算的最小二乘定位思想；（2）针对非圆信号的直接定位问题，设计基于加权扩展子空间数据融合的多维参数联合直接定位算法与基于神经计算的误差校正方法，兼顾直接定位性能与计算能力；（3）研究非圆信号直接定位算法的性能指标与定量分析方法。项目致力于融合利用多维参数信息，丰富非圆信号无源定位的理论与方法，为提高现有非圆信号定位技术水平提供新的理论基础与技术支撑。

联合多维参数的定位方法能够获取丰富的定位信息，有助于为工业生产和军事应用提供准确的目标位置，在定位中利用非圆信号椭圆协方差不为零的特性，能够进一步提升定位性能。为此，项目从两步定位与直接定位两种方式，围绕多维参数联合的两步定位理论与方法、面向非圆信号的直接定位理论与方法展开研究，其中考虑了站址误差、卫星星历误差以及阵列互耦误差等多种误差存在的定位场景，并定量分析了不同误差影响下定位算法的理论性能。项目的研究成果为提高现有无源定位技术水平提供了新的理论基础与技术支撑。项目的主要研究工作和创新点总结如下：.（1）针对多维参数联合的两步定位问题，提出了基于加权多维标度和拉格朗日乘子的运动源到达时间差(TDOA, Time Difference of Arrival)和到达频率差(FDOA, Frequency Difference of Arrival)定位算法、基于加权多维标度分析的多个非相关源TDOA/FDOA协同定位算法、协同陆基与星基系统的超视距目标最小二乘定位算法、以及针对超视距辐射源的陆基测角与测时差协同定位算法。项目通过一阶误差分析证明，所提方法对辐射源位置（速度）的估计精度均能渐近逼近相应的克拉美罗界。仿真结果显示所提方法具有渐近最优性能，与基于单一参数的定位算法相比，多维参数联合的两步定位算法具有更高的误差阈值和更强的抗噪性能。.（2）针对面向非圆信号的直接定位问题，提出了卫星星历误差条件下的非圆信号直接定位算法、方位依赖幅相误差影响下的非圆信号多站直接定位算法、面向非圆信号的四阶累积量直接定位算法、互耦误差影响下的非圆信号直接定位算法、以及模型误差条件下基于卷积神经网络(CNN, Convolutional Neural Network)+双向长短时记忆网络(BiLSTM, Bi-directional Long-Short Term Memory)的非圆信号目标直接跟踪算法。项目定量分析了卫星星历误差、阵列互耦误差影响下非圆信号直接定位算法的理论性能。仿真结果显示所提算法在上述各种误差条件下具有较强的鲁棒性能，且与传统算法相比，利用信号非圆特性的直接定位算法具有更强的空间分辨能力与更高的定位精度。

内容获取失败，请点击重试