ID	原文	译文
7914	为了对故障预测的研究有一个系统、全面的认识,本文从故障预测的内容、对象、方法3个维度对基于状态的武器电子装备故障预测研究进行了综述,并对相关的故障预测方法进行了分析和梳理。	In order to study on failure prediction has a systematic and comprehensive understanding, this article from the content of the fault prediction, object, method of three dimension of weapons based on state of the electronic equipment fault prediction research is summarized, and the relevant fault prediction methods are analyzed and combed.
7915	针对装备故障预测问题,立足于文献综述,提出一种基于状态的装备故障预测分析方法,为部队对装备进行合理、科学的故障预测提供借鉴。	According to equipment failure prediction problem, based on the present review, in this paper, a fault forecast analysis method based on state of the equipment, equipped with forces of the rational and scientific reference for failure prediction.
7916	针对大规模监测系统中经常出现的监测点失效、数据异常等问题,提出基于深度降噪自编码网络的监测数据修复方法。	Aiming at monitoring stations often appear in large scale monitoring system failure and abnormal data, noise reduction is put forward based on depth from coding network monitoring data repair methods.
7917	首先,通过堆叠降噪自编码构造深度降噪自编码网络来提取监测点之间隐含的深层关联关系,进而,基于这种深层关联关系训练一种支持向量回归模型以预测待修复的监测数据。	First, by the stacked noise from noise reduction since coding network coding structure depth to extract implied the profound relationship between monitoring stations, in turn, based on the deep relationship training a support vector regression model to predict the monitoring data for repair.
7918	在某机场噪声实测数据上的实验表明,通过深度降噪自编码网络学到的深层关联关系能够有效地重构噪声监测数据;	At a airport noise on the measured data of the experimental results show that through deep noise from the coding network learned the deep relationship can effectively reconstruct noise monitoring data;
7919	相比传统数据修复方法,所提出的数据修复方法具有更好的鲁棒性,数据的修复具有更高的精度。	Compared with traditional data repair methods, the proposed data repair method has better robustness, repair has a higher accuracy of the data.
7920	为了进一步拓宽度量学习在图像分类中的适用范围,同时提高分类的性能,本文提出一种基于椭圆-双曲线马氏度量的图像分类算法。	In order to further broaden the measurement study scope in the image classification, at the same time improve the performance of classification, this paper proposes a image classification algorithm based on elliptic - hyperbolic markov measure.
7921	该算法首先将颜色特征和局部二值模式描述的纹理特征相结合来表示图像特征;	The algorithm firstly described characteristics and local binary pattern color texture features to represent the combination of image feature;
7922	然后引入对样本数据具有更好的适应性的椭圆-双曲线度量,根据数据统计特性定义椭圆-双曲线马氏度量,给出椭圆-双曲线马氏度量学习算法,从而获取最优的度量矩阵;	Then introduced to the sample data has better adaptability of elliptic - hyperbolic measure, according to the data statistical properties define elliptic - hyperbolic markov measures, elliptic, hyperbolic markov measure learning algorithm are given, so as to obtain the optimal measure matrix;
7923	最后利用椭圆-双曲线马氏度量矩阵将样本变换到新的特征空间,从而降低特征各维度间的相关性,同时计算图像特征间的距离从而完成分类。	Last samples using elliptic hyperbolic markov measure matrix will transform to the new feature space, thus reduce the correlation between the characteristics of each dimension, and calculate the distance between image features and classification.