In order to solve the problem of an uncertain initial state and big errors for hypersonic glide vehicle (HGV) tracking, a hybrid model algorithm is proposed by combining a single model algorithm with a multiple model algorithm. To develop the tracking algorithm with the Cubature Kalman filter, in every model filter the process equation is established based on the HGV aerodynamic model and the measurement equation is established based on the radar measurement principle. The proposed hybrid model algorithm is developed by using the multiple model algorithm in the initial tracking stage and using the single model algorithm in the stable tracking stage, and they are divided by a proposed parameter. The former can avoid divergence and reduce the errors caused by the uncertain initial state. The latter can track the HGV at higher accuracy. The simulation indicates that the proposed hybrid model has high speed accuracy in the whole tracking stage and high position accuracy in the stable tracking stage. The average position root mean square error (RMSE) using the hybrid model algorithm is almost the same as that using the single model algorithm but the average speed RMSE using the single model algorithm is about 30% greater than that using the hybrid model algorithm. In a system for defending the HGV, the speed accuracy has more effect on the trajectory prediction as time goes on. Thus, the hybrid model algorithm is an engineering algorithm for HGVs with high accuracy. In future research, the hybrid model algorithm will be studied for general maneuvering target tracking.
1 November 2022 in Journal of Physics: Conference Series
In recent years, three-dimensional convolutional neural network (3D CNN) has been widely used in the fields of action recognition and video analysis. The general purpose processors are difficult to achieve efficient and intensive computing, and the deployment of 3D CNN based on FPGA has the advantages of low power consumption, high energy efficiency, and customizability, and has gradually become a hot choice for deploying convolutional neural networks in many embedded scenarios. This paper designs a small 3D convolutional neural network based on the classic 3D convolutional neural network C3D, and uses the general matrix multiplication (GEMM) to map the 3D convolution calculation to the 2D matrix multiplication calculation. The matrix is divided into blocks and transmitted to the FPGA through the AXI bus, and the multiplication operation of the block matrix is realized through a two-dimensional multiply-accumulate array. A System on Chip (SoC) architecture is built on the PYNQ platform, using ARM Cortex-A9 as the process control core, and the calculation of the entire matrix is completed under reasonable block and scheduling on the ARM. The IP core of matrix calculation is designed using High Level Synthesis (HLS), and the corresponding parallel optimization scheme is given. Experiments have verified that the prototype design of the hardware acceleration circuit achieves low power consumption, high energy efficiency, and high precision motion recognition while using less hardware resources.