基于动态变结构BP神经网络的目标威胁评估

针对传统目标威胁估计方法和BP神经网络的不足,在BP神经网络的基础上,建立了基于动态变结构BP神经网络的目标威胁估计模型.该模型通过在权值向量更新公式中引入冲量函数,加快了网络的搜索速度和精度,保证了网络获得全局最优值;通过实时调整隐含层节点数目,可以将网络结构优化,极大地提升了网络的灵活性.仿真结果表明,与传统目标威胁估计方法和BP神经网络相比,动态变结构BP神经网络具有更好的预测能力和收敛速度,可以快速、准确地完成目标威胁估计.

Evaluation of target threat assessment based on dynamic structure changed BP neural network

For the defects of the traditional target threat estimation method and Back-Propagation (BP) neural network,based on the introduction of BP neural network,a target threat assessment model is proposed.By introducing a momentum function into weight vector update formula,the accuracy and search speed of the network were improved,and the optimal value was guaranteed.By adjusting the number of the hidden nodes in real time,the structure of the network was optimized,thus,promoting the network's flexibility.The simulation results show that,compared with the traditional target threat estimation method and BP neural network,dynamic structure changed BP neural network has better prediction ability and can estimate air target threat quickly and accurately.