| 摘要: |
| 针对多雷达数据融合问题，提出了基于时间序列的聚类算法，用于实现航迹相关，即以时间序列为基础把聚类模型转化为基于特征匹配的聚类算法。进一步考虑到多目标密集时，部分来自不同目标的数据可能比来自同一目标的数据更接近，易导致关联错误，为此提出了基于时间序列的模糊聚类算法。对上述两种算法的聚类结果，应用卡尔曼滤波器实现滤波跟踪，在不同的情况下仿真后发现，在跟踪目标较少且相互位置较远的情况下，两种算法均有效，在跟踪目标较多且相互位置靠近的情况下，基于时间序列的模糊聚类算法更有效。 |
| 关键词: 雷达数据融合 模糊聚类 时间序列 特征匹配 |
| DOI： |
|
| 基金项目: |
|
| Multi-radar data fusion based on time series clustering |
| ZHANG Qianjun |
| (Southwest China Institute of Electronic Technology,Chengdu 610036,China) |
| Abstract: |
| A clustering algorithm based on time series is proposed to realize track correlation in multi-radar data fusion,which transforms the clustering model based on time series to a clustering algorithm based on feature matching.In intensive multi-objective situation,that some data from different targets may be closer than the data from the same target may lead to association errors,so a fuzzy clustering algorithm based on time series is proposed.Two algorithms in different simulation cases are found that when objectives are few and far between each other,both the two algorithms are effective,and when objectives are intensive and close,the latter algorithm is more efficient. |
| Key words: radar data fusion fuzzy clustering time series feature matching |
