事件驱动采样长路径多源数据的快速跟踪与溯源方法研究

项目名称： 事件驱动采样长路径多源数据的快速跟踪与溯源方法研究

项目编号： No.61273002

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 金学波

作者单位： 北京工商大学

项目金额： 61万元

中文摘要： 含有事件驱动采样长路径多源数据的物联网RFID目标跟踪系统是物联网最重要的组成部分，有着重要的应用价值。按照时间顺序逐一递推的传统融合估计方法无法保证系统跟踪与溯源的实时性，而利用数据关系模型快速查找目标移动序列、并以此序列近似地表示目标移动路线的方法精确度不高。项目研究即有效融合长路径多源数据、又可根据系统特性进行高速运算的估计理论与算法，内容包括1）研究基于目标运动特性的动态分簇方法，将不规则测量时刻的融合估计问题描述为贝叶斯框架下的估计问题；2）研究当前状态与簇中成员数据引起的测量失序数据的数学关系，得到基于Kalman滤波器、粒子滤波器及MHE算法的渐进融合估计算法；3）研究物联网RFID系统模型参数与状态估计间的数据交换机制，研究参数及状态并行估计理论与算法。成果将从根本上解决由事件驱动采样长路径数据引起的估计性能下降、实时性差的问题，在理论研究与实际应用中都具有极其重要的价值。

中文关键词： RFID目标跟踪系统；自适应过程模型；不规则采样理论；弹簧跳采样模型；非线性跟踪

英文摘要： As the most important part of the Internet of Things (IoT), Radio Frequency Identification (RFID) tracking system based on long path multi-source data with event triggered sampling has the important applications. The traditional fusion method with recursive one by one can't get real-time estimation, while the estimation get by moving target sequence method based on searching data relationship model is not accurate enough. Research project here focus on fusion estimation theory and algorithms that can deal with long path multi-source data, meanwhile, obtain high-speed computation in accordance with the feature of the system. The main works will include 1) to study dynamic clustering method based on target motion characteristics, and describe the irregular sampling estimation problem as the estimation problem within the Bayesian framework; 2) to study the mathematical relationship of the current state and the out-of-sequence measurement caused by members in the cluster data, then get progressive fusion estimation algorithm based on Kalman filter, particle filter and MHE, etc; 3) to study data exchange mechanism of model parameters and state estimation in RFID of IoT system, and get the parallel theory and algorithms with model parameters and state estimation. Project results will fundamentally resolve the performa

英文关键词： RFID tracking system；the adaptive process model；irregular sampling theory；the spring jump sampling model theory；nonlinear tracking