Abstract:

Aiming at the problem of low accuracy caused by complex electromagnetic environments, high track density, and insufficient extraction of track information in the current multi-sensor track correlation algorithm, the track correlation problem characterizes as a classification problem in the field of deep learning, and an unmanned aerial vehicle target track correlation method is proposed based on graph convolutional network （GCN）. By converting the target data detected by sensors with different measurement errors into graph data, using correlation pair processing, and inputting it into GCN for feature extraction, the correct track correlation is realized. The simulation results show that under the same simulation conditions, the proposed method is superior to the contrast algorithms, which can fully extract the high-dimensional features of the track and take into account the previous track information, and has good research prospects and research value.

Key words: track correlation, deep learning, machine learning, graph convolutional network （GCN）