ID	原文	译文
6574	针对红外场景仿真效果评估困难的问题,提出了基于自学习框架的评价方法。	According to infrared scene simulation evaluation difficult problem, put forward the evaluation method based on self learning framework.
6575	首先,从仿真图像与实际图像视觉对比的角度提出了面向图像视觉相似的红外场景仿真效果评价指标体系,用于量化评价过程;	First of all, from the perspective of visual contrast simulation images and real images in image was proposed and similar visual effect evaluation index system of the infrared signature scene simulation, is used for quantitative evaluation process;
6576	其次,提出以极限学习机(extreme learning machine,ELM)为核心建立评估模型,建立包括蒙特卡罗样本仿真、ELM评估网络及自更新仿真样本评估模型等3部分在内的自学习框架来生成仿真样本、强化对ELM的训练;	Secondly, the paper puts forward to extreme learning machine (extreme learning machine, ELM) as the core to establish the evaluation model, set up including samples of monte carlo simulation, ELM evaluation network simulation sample evaluation model and the update of three parts, such as self learning framework to generate simulation sample and to strengthen the training of ELM;
6577	最后,针对实际样本数量较少的问题,在此框架基础上提出了包括样本评定、自学习、评估模型测试3个阶段在内的仿真图像相似性评估方法,实现了从样本生成到评估过程的自动化。	Finally, aiming at the problem of less actual sample size, on the basis of this framework is proposed including sample evaluation, self learning, evaluation model test of three stages, including the simulation image similarity assessment method, automate the generated from the sample to the evaluation process.
6578	实验结果表明提出的自学习框架能够显著提高评估模型的正确率,而且训练后的评估模型适用性强,可独立自主进行红外场景仿真效果评估。	The experimental results show that the proposed self-learning framework can significantly improve the accuracy of assessment model, and after the training evaluation model applicability, can be an independent infrared scene simulation evaluation.
6579	针对网络控制系统(networked control system,NCS)诱导时延具有的时变、随机、非线性等特点,提出了一种相空间重构与鲁棒极限学习机(robust extreme learning machine,RELM)的时延预测算法。	For networked control systems (networked control system, and replication) induced delay has the characteristics of time-varying, random, nonlinear, puts forward a kind of phase space reconstruction and robust extreme learning machine (robust extreme learning machine, RELM) delay prediction algorithm.
6580	首先利用0-1测试对时延序列进行混沌特性检测,再通过改进关联积分法确定重构延迟参数和嵌入维数,进而对时延序列进行重构,新的样本更能真实反映时延变化特性。	First using 0-1 test to delay sequence is used to detect the chaotic characteristics, and by improving the correlation integral method refactoring delay parameters and embedding dimension, and then to delay sequence of refactorings, new samples can reflect more delay variation characteristics.
6581	以重构后的时延序列为训练样本,同时,考虑异常值的稀疏特性,运用RELM进行时延序列预测。	The sequence of time delay reconstruction as the training sample, at the same time, considering the sparse characteristic of outliers, using RELM delay sequence prediction.
6582	该方法具有学习速度快、泛化性能好、可有效降低异常值影响等优点。	This method has fast learning speed, good generalization performance, can effectively reduce the abnormal value influence, etc.
6583	信源数估计的性能直接影响着高分辨测向的精度。	Source number estimation performance directly affects the accuracy of high resolution direction finding.