基于 BP 神经网络的自适应偏置比例导引

To address the drawback of traditional analytical biased proportional guidance with poor guidance accuracy when maneuvering in a wide range,an adaptive biased proportional guidance law based on BP(Back propagation) neural network is proposed. The bias term is accurately solved online through the BP neural network. Firstly,the error of solving bias term in analytic form is investigated. Specifically, the influence of different parameters on the solution error of bias term is demonstrated. Secondly,the mapping relationship between parameter and constant term is proved. BP neural network is used to fit the mapping accurately. Thirdly,sensitivity analysis was performed for multidimensional input parameters,on this basis,equilibrium samples for BP neural network in parameter space batch are generated. Finally,the bias term solution model based on BP neural network is established and Adam learning method is used to train the network. In addition,the stability of the guidance law is proved in theory. The effectiveness of the training is tested and verified by mathematical simulation. The simulation results show that the proposed method can be implemented with limited computational cost and effectively improve guidance accuracy,and the average impact angle error is 0. 024毅. This paper provides a reference for engineering application.