面向弹道导弹防御的快速最优拦截弹道规划

A rapid optimal intercept trajectory planning method based on neural network is proposed for solving the intercept trajectory planning problem of ballistic missile defense. Firstly,the optimal intercept trajectory cluster is generated by optimizing the trajectory of the powered phase. Then,the optimal intercept point and intercept trajectory with respect to the maximum intercept velocity are determined based on the optimal intercept trajectory cluster. To improve the efficiency,the neural networks are used to establish the mapping relationships between the intercept point and the intercept velocity,intercept time,and trajectory parameters. For the trajectory optimization problem of the powered phase,a trajectory optimization algorithm based on nonlinearity-kept convexification is proposed. This paper integrate the nonlinearity of dynamics by using the Taylor formula,properly preserve the nonlinearity by defining new variables and introducing new constraints,and then linearize an obtained control-affine system to get linear dynamics. This method effectively improves the convergence of the algorithm,and the convergence issue caused by directly linearizing dynamics can be avoided. Simulation results demonstrate that the proposed method has good reliability and real-time performance.