Application of Deep neural Network in Air Target Threat Assessment

摘要

Air target threat assessment is a key factor in airborne sensor resource management. In order to improve the accuracy of evaluation, this paper proposes a new method of target threat assessment based on Rectified Linear Units (ReLU) deep neural network (RL-DNN). Target threat assessment model and algorithm for RL-DNN predictor are established. The model adds the number of network layers based on the BP neural network, using ReLU as the activation function, and the Adam algorithm for back propagation, which effectively avoiding the problem of slow learning rate and falling into local extremum. Experimental results show that the algorithm based on RL-DNN is higher than that based on the traditional algorithms, which proves the proposed algorithm has good predictive ability and can accurately complete the air target threat assessment.