ID	原文	译文
2153	针对短波信道下信号截获质量差，信道环境复杂以及单一特征识别率低等问题，提出了基于深度残差网络的信号特征自动提取算法，设计了一种具有自适应学习能力的短波特定通信协议识别模型。	To correctly classify the specific protocol signal, a signal recognition model with adaptive learning and au-tomatic feature extraction ability is proposed. This model is based on the deep residual network, which can solve the draw-backs, such as, the poor quality of the intercepted communication signal, the complex condition of the short wave channel, and the low recognition rate of the single feature.
2154	通过对具有特殊结构的协议信号的时频视觉差异进行理论推导，将信号的时频能量转换成灰度图像，并用于对所构建的深度残差网络进行训练。	After analyzing the visual characteristic of communication protocol signal with special structure in time frequency domain, the time frequency gray-images are obtained and utilized to train the deepresidual network. This method does not need much prior knowledge and is insensitive to signal quality.
2155	该方法克服了传统方法对信号质量要求高、先验信息需求多等缺陷，可直接对中频接收信号进行处理，适合实际工程应用。	Moreover, it can process the intermediate-frequency signal directly. Due to these advantages, the algorithm is suitable for engineering applica-tion.
2156	实验表明，当深度残差网络达到稳态时，识别准确率高，在低信噪比、多径衰落、多普勒频偏以及信号被强干扰所遮挡的情况下，依旧能准确识别协议类别。	Simulation results show that, when the deep residual network reaches its steady status, the proposed model can accurate-ly identify the protocol. And it is also proved effective even at complex circumstance where the multipath fading and theDoppler shift exist, the signal-to-noise ratio is low, and the interference is strong.
2157	为了提高非线性卫星钟差预测的精度，降低单一钟差预测模型对钟差预测的风险，提出了一种组合模型的卫星钟差预测算法。	In order to improve the accuracy of nonlinear satellite clock bias prediction and reduce the risk of singleclock bias prediction model for clock bias prediction, a satellite clock bias prediction algorithm based on combined model is proposed.
2158	该算法首先采用 db1 小波对卫星钟差序列进行 3 层多分辨率分解和单支重构，得到一个趋势分量和三个细节分量，	The algorithm firstly uses db1 wavelet to conduct a 3 layer multiresolution decomposition and single branch recon-struction of satellite clock bias sequences, and obtains a trend component and three detail components.
2159	然后运用灰色预测模型对重构后的趋势分量和混沌一阶加权局域预测法对重构后的细节分量分别进行预测，	Then the grey predic-tion model is used to predict the reconstructed trend component and the chaotic time series one-order weighted local predic-tion method is used to predict the reconstructed detail components.
2160	最后将各分量预测结果相加后得到总的钟差预测值。	Finally, each component prediction result is added to ob-tain the total clock bias prediction value.
2161	以 GPS 卫星钟差数据做算例分析，在 6 小时的钟差预测中，算法绝对误差最大值比单一的灰色预测模型误差小 1.3ns 以上。	Taking the GPS satellite clock bias data as an example, the maximum absolute errorof the algorithm is at least 1. 3ns smaller than that of a single gray prediction model in 6-hour clock bias prediction.
2162	将该组合预测模型用于非线性卫星钟差预测中，可以提高钟差预测的精度和可靠性。	This combined prediction model can be applied to the prediction of nonlinear satellite clock bias, which can improve the accuracyand reliability of clock bias prediction.