A Radar/Telemetry Information Fusion Method Based on Tcn-Iukf

In ballistic measurement, sensors from different platforms are used to improve accuracy. However, due to the differences in measurement characteristics of the sensors, traditional Kalman filtering methods struggle to fully utilize multisource information. Especially in ballistic motion, the complex motion equations and significant interference make it difficult to accurately establish the state transition equation, leading to low fusion accuracy. Therefore, this paper proposes a ballistic fusion algorithm based on Time Convolution Network Improved Unscented Kalman Filter (TCN-IUKF). First, the state models for the missile-borne IMU+GNSS integrated navigation and the ballistic tracking radar are established using TCN. Then, the trained TCN model is used to replace the state and observation models in the UKF, and an equidistant sigma sampling method is applied. Finally, experimental results demonstrate the effectiveness of the ballistic fusion algorithm. The results show that the TCN-IUKF algorithm reduces the root mean square errors in three-axis position by$84.5 \%, 81.6 \%$, and 55.3 %, respectively, compared to existing algorithms.