基于梯度下降法的离散概率空间区域射击瞄准点求解方法

Gradient Descent-based Aiming Point Solving Method for Area Firing in Discrete Probability Space

为增强区域射击方法的适用性,解决目标分布不规则条件下的区域射击瞄准点求解问题,提出基于梯度下降法的离散概率空间区域射击瞄准点求解方法。通过在离散空间内描述目标的分布概率,以适应不规则的目标分布情况;通过变分方法求解离散空间内的最优中间函数,以实际中间函数与最优中间函数的差值范数作为优化指标,给出解析形式的指标梯度;构建基于梯度下降法的射击点位置求解模型。在同等初始条件下,分别采用几何法、函数逼近法、梯度下降法求解得到不同目标分布情况下的瞄准点配置,以对比方法性能。结果表明:相比于现有方法,梯度下降法适用性强、求解速度较快、优化程度高,能够求解目标分布不规则情况的区域射击瞄准点配置。

A gradient descent-based aiming point solving method for area firing in discrete probability space is presented to enhance the applicability of area firing method and solve the area firing aim points in the case of irregular distribution of targets.The target distribution probability in discrete probability space is described to adapt to irregular target distribution.The optimal intermediate function in discrete probability space is deduced using the variational method.The analytical gradient function of index is deduced by setting the norm of difference value between practical intermediate function and optimal intermediate function as index function.An aim point solving model is constructed based on gradient descent method.The geometric method,function-approximation method and gradient descent method are used to obtain the distributions of aiming points under the equivalent initial condition for different target distributions,respectively.The results show that the gradient descent method has better applicability,faster solving speed and higher degree of optimization for solving the aim-points configuration for area firing under the condition of irregular target distribution.