具有噪声和传输不确定性的多传感器系统融合滤波方法

项目名称： 具有噪声和传输不确定性的多传感器系统融合滤波方法

项目编号： No.61304258

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 冯肖亮

作者单位： 河南工业大学

项目金额： 25万元

中文摘要： 现有的噪声不确定多传感器系统融合滤波方法大都借助于高维矩阵运算处理，计算负荷高。而非理想信道条件下的传输不确定性，又使得信息处理时会增加新的噪声不确定性。现有的多传感器延迟量测融合滤波方法大都存在滤波场景受限、实时性差等问题。本项目结合信息融合理论、凸优化理论及Krein空间理论，拟开展以下研究旨在建立实用、实时、低维的融合滤波方法以增强网络感知系统的信息处理能力：1）研究递归正交变换方法，在集中式与分布式框架下分别建立包含噪声有限步相关性系统的低维融合滤波理论。噪声统计特征未知时，开展不同性能指标约束下的序贯式H∞融合滤波研究。2）开展非理想信道条件下全过程性能最优的序贯式融合滤波问题研究。并建立延迟量测选择机制，研究具有更高性价比的延迟选择驱动序贯式融合滤波方法。3）建立同时包含传输不确定性、能量有界噪声不确定性多传感器系统的性能指标函数，给出全过程性能最优的序贯式H∞融合滤波机制。

中文关键词： 噪声不确定性；传输不确定性；延迟量测选择机制；Krein 空间；序贯式融合 滤波

英文摘要： The existing fusion filtering methods for multi-sensor systems with noise uncertainties are mostly deduced based on high dimension matrix operations, with high computation loads. New noise uncertainties will be increased, when processing with communication uncertainies in non-ideal communication channels. There are several problems with the existing fusion filtering methods for multi-sensor systems with delay measurements, such as constrained filtering scenes, poor real-time performance. By means of the information fusion theory, convex optimization theory, Krein space theory, several practical, real-time, lower dimensional fusion filtering methods will be deduced in this project, in order to enhance the information processing ability for networked acquisition systems. 1) Research the recursive orthogonal transformation method, and built the lower dimensional fusion filtering methods for the systems with finite-step noise correlations. When noise statistical characteristics are unknown, the sequential H∞ fusion filtering methods will be proposed under different performance indexes. 2) Research the sequential fusion filtering problem for the global filtering process optimal, in non-ideal communication channels. Built the delay measurement selection mechanism, driven by which several sequential fusion methods will

英文关键词： Noise Uncertainty；Communication Uncertainty；Delay measurement selection mechanism；Krein Space；Sequential Fusion Filtering