Track fusion algorithm for OTHR network based on deep learning

This paper presents a deep learning based over the horizon radar (OTHR) track fusion method, which fuses the tracks of two OTHRs to improve the track accuracy of the OTHR fusion system. This method generates random track data of two OTHRs within a certain distance and azimuth range through simulation, forming data samples. Firstly, at the fusion center, preprocess the local estimation results of the two OTHRs through methods such as time alignment and coordinate transformation. Then divide the dataset of the track data of the two OTHRs into samples to obtain training and testing samples. Secondly, we constructed a deep neural network model and trained the deep neural network using training sample data. Finally, the deep neural network is tested using the test trajectory data, and the fused trajectory is outputted by the network. The proposed method fully utilizes the powerful learning ability of deep learning technology and does not require a single OTHR to provide covariance information, making the track fusion system easier to implement. Due to the fact that the deep track fusion method proposed in this article only requires one forward propagation in practical use, it has a smaller computational complexity. Experimental results show that the track fusion algorithm proposed in this paper has higher fusion accuracy and fusion efficiency than the traditional simple convex combination and covariance cross fusion methods, which verifies the effectiveness of the proposed method.