ID 原文 译文
14595 目前任务系统综合化程度不断提高,故障诊断难度不断加大。 At present,the integration degree of mission system is constantly improving.
14596 结合任务系统的使用场景以及综合化系统的结构特点,对PHM系统的层次结构、基础资源要素、功能要素、综合诊断、状态策略、故障管理等环节开展了全面设计与实现。 Considering the application scenarios of the mission system and the structural characteristics of the integrated system,this paper designs and implements the hierarchical structure,basic resource elements,functional elements,comprehensive diagnosis,state strategy and fault management ofPHM system.
14597 设计结果表明,PHM系统在故障诊断准确性、降低飞机再次出动执行任务的间隔时间、缩短地面维修时间方面均具有明显优势。 The results show that the PHM system has obvious advantages in the fault diagnosis accuracy,reducing the interval time of aircraft to launch again and shortening the ground maintenance time.
14598 传统雷达高分辨一维距离像(High-resolution Range Profile,HRRP)目标识别方法只利用目标幅度信息而丢失其相位信息,这势必会造成信息不完备。 The traditional radar high-resolution range profile(HRRP) target recognition method only usesthe target amplitude information and loses its phase information,which will inevitably cause incomplete information.
14599 为解决此问题,提出将深度极限学习机从实数域扩展到复数域,以有效提取复HRRP序列的深层潜在结构信息。 In order to solve this problem,this paper proposes to extend the deep extreme learning machinefrom the real domain to the complex domain to effectively extract the deep potential structural information ofthe complex HRRP sequence.
14600 同时为更好地保持数据间的邻域信息,将流形正则化引入到网络模型训练过程中,提出流形正则深度复极限学习机。 At the same time,in order to better maintain the neighborhood relationshipsbetween data,manifold regularization is introduced into the training process of network model,and the manifold regularization complex deep extreme learning machine is proposed.
14601 在雷达暗室测量数据上的实验结果表明,所提算法相比常用的深度学习模型具有更好的识别效果和更快的训练速度,验证了算法的有效性。 Experimental results on radar darkroom measurement data show that the proposed algorithm has better recognition effect and faster trainingspeed than commonly-used deep learning models,which verifies the effectiveness of the algorithm
14602 当前先进的图像检索方法中,存在着不能很好地分辨图像中不同区域和内容的重要性的问题,导致计算资源分配不合理、检索正确率较低等一系列结果。 The importance of different regions and contents in the image cannot be well distinguished in thecurrent advanced image retrieval methods,which leads to the results like unreasonable allocation of computational resources and low accuracy.
14603 为了解决这些问题,提出了一种基于卷积神经网络(Convolutional Neural Network,CNN)和注意力机制的图像检索方法。 To solve these problems,this paper proposes an image retrieval methodbased on convolutional neural network(CNN) and attention mechanism.
14604 首先使用卷积神经网络提取特征,然后使用注意力机制处理提取的特征,可以在计算能力有限的情况下根据图像中的内容合理分配计算资源,使图像中的突出部分得到更多的关注。 First, the CNN is used to extractfeatures. Then the attention mechanism is used to process the extracted features,which can reasonably allocate computing resources according to the contents of the image in the case of limited computing power,sothat the highlights in the image can get more attention.