基于智能化蚁群混合行为的数据关联技术研究

项目名称： 基于智能化蚁群混合行为的数据关联技术研究

项目编号： No.60804068

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 徐本连

作者单位： 常熟理工学院

项目金额： 14万元

中文摘要： 密集杂波下的纯方位多传感器多目标定位与跟踪问题广泛应用于海、陆、空等军事或民用航空领域，当传感器的数目增加时，其对应的数据关联问题的求解是NP的，近30年来它一直倍受国内外学者的关注。由于蚁群算法是解决NP问题的一种有效手段，并且，问题的求解空间也很适合用蚁群算法进行研究，因此，本项目拟把智能化的蚁群混合行为引入到该领域，以期建立一套基于蚁群算法的多目标数据关联方法和参数估计方法，提出多任务蚂蚁来解决多目标航迹的起始问题（航迹数目已知）；提出基于三原色蚂蚁解决目标的检测和数据关联问题；把多目标数据关联问题转化成聚类和分类问题，分别建立其基于蚂蚁的航迹起始规则库，实现回波的聚类和分类；最后，基于粒子滤波器（权重调整）思想，应用蚂蚁来近似目标状态的后验概率密度函数，继而提出一种蚂蚁滤波器。

中文关键词： 蚁群算法;参数估计;数据关联;多目标跟踪

英文摘要： The problem of multi-sensor-multi-target bearings-only tracking in a dense and cluttered environment is widely considered in various fields, such as in a military/civilian sea, land,and air field. With the increase of the number of sensors, the corresponding complexity of data association becomes an NP problem, thus it has been attracting many researchers’ttentions in this field over the last 30 years. Due to the fact that the ACO is recognized as an effective means to solve various combinatorial optimization problems,it is natural to introduce the intelligent hybrid behaviors of ant colony into the area of data association and the area of parameter estimation. To solve the data association problem, three types of techniques are presented. First, an algorithm of ants with different tasks is investigated given the known number of targets.Second,a three-primary-color-ant-based technique is proposed when the number of targets is unknown. Third, a set of rules based on the ACO are constructed and employed to classify or cluster the obtained returns. Additionally, inspired from particle filter, several novel parameter algorithms, called ant estimators, are designed to estimate the posterior probability density functions of interested states.

英文关键词： Ant colony optimization; Parameter estimation; Data association; Multi-target tracking