基于频带相关性Deep Learning的无线通信干扰智能识别

近年来,随着无线通信技术在军事领域的发展及各类无线通信装备数量的增加,战场电磁环境变得越来越复杂,其要求通信系统具有更强的抗干扰能力,而智能干扰识别是抗干扰的前提。基于传统机器学习的各类干扰识别算法存在干扰前期特征提取繁杂、低干噪比(JNR)下识别正确识别率低的问题。本文针对上述问题,引入了并行多路多尺度卷积,通过增加网络宽度、提升网络识别速度和精度,解决前期特征提取繁杂的问题;结合频带相关性长短时记忆网络(LSTM)创新性地提出了基于LSTM的频带相关性无线通信干扰智能识别网络,该方法利用LSTM探知干扰信号频带相关性,提升低干噪比下的干扰识别正确率。通过仿真证明了多尺度并行卷积比普通卷积网络更稳定,识别速度和泛化性更佳,该网络提升了低干噪比下的识别正确率,平均识别正确率达99.86%。可见,本文所提的网络模型是解决复杂电磁环境下的智能干扰识别的一种有效可用模型。

Investigation of the two-quasiparticle bands in the doubly-odd nucleus ^(166)Ta using a particle-number conserving cranked shell model

The high-spin rotational properties of two-quasiparticle bands in the doubly-odd 166Ta are analyzed using the cranked shell model with pairing correlations treated by a particle-number conserving method, in which the blocking effects are taken into account exactly. The experimental moments of inertia and alignments and their variations with the rotational frequency hw are reproduced very well by the particle-number conserving calculations, which provides a reliable support to the configuration assignments in previous works for these bands. The backbendings in these two-quasiparticle bands are analyzed by the calculated occupation probabilities and the contributions of each orbital to the total angular momentum alignments. The moments of inertia and alignments for the Gallagher-Moszkowski partners of these observed two-quasiparticle rotational bands are also predicted.