ID	原文	译文
55707	在不考虑系统模型误差的情况下,为降低设计的保守性,利用描述残差对故障灵敏度的H-指数和表示残差对未知输入鲁棒性的H∞范数以及一种新的迭代线性矩阵不等式(linear matrix inequality,LMI)算法解决H∞/H-的优化问题,设计了参考残差模型和自适应门限,并将观测器设计转化为LMI表述的标准H∞模型匹配问题,减小故障检测过程中误报、漏报率。	Under the condition of without considering the system model error, to reduce the conservative design, using description in residual fault residual sensitivity index of H - and said to the unknown input robust h-infinity norm and a new iterative linear matrix inequality (LMI) (linear matrix inequality, a LMI) algorithm to solve the problem of the optimization of the h-infinity/H -, design reference for residual error model and adaptive threshold, and the observer design into LMI representation of h-infinity model matching problem, reduce false positives, non-response rates in the process of fault detection.
55708	仿真结果表明,设计的故障检测观测器对故障的灵敏度高,对不确定因素的鲁棒性强。	Simulation results show that the design of fault detection observer for fault of high sensitivity, strong robustness to uncertainties.
55709	为了进一步提高超分辨图像重建效果,针对前期研究的超分辨误差反向传播神经网络(back propagation neural network,BPNN)重建结果中存在的块痕迹问题加以改进和优化。	In order to further improve the effect of super resolution image reconstruction, in view of the previous studies of super resolution error back propagation neural network (back propagation neural network, the BPNN) block traces problems existing in the reconstruction result is improved and optimized.
55710	对影响BPNN超分辨效果的两个关键问题进行改进:(1)网络训练样本问题,将8×8→16×16的映射方式改进为2×2→4×4的映射方式,同时,采用相邻仅间隔一个像素的方式优化构造训练样本;	Affecting the BPNN super resolution to improve the effect of the two key problems:
55711	(2)加速网络训练收敛问题,将网络训练规则由BP算法改进为改进的比例共轭梯度算法。	(1) network training samples, the 8 x 8 to 16 x 16 mapping way of improvement of the 2 x 2-4 x 4 mapping way, at the same time, adopt the way of adjacent only one pixel interval optimization structure training samples;(2) a convergence speed up network training, network training rules by the improved BP algorithm to improve the proportion of the conjugate gradient algorithm.
55712	网络训练实验和泛化实验表明,改进方法增加了网络训练样本数量,改善了超分辨BPNN的输出图像质量,有效解决了超分辨结果中的块痕迹问题,使超分辨结果图像的峰值信噪比提高约8dB。	The network training and generalization experiments indicates that the improved method increases the number of training samples, improve the quality of the output image super resolution BPNN, effectively solve the problem of block traces in super resolution results, make the results of super resolution image peak signal-to-noise ratio increased by 8 db.
55713	提出一种双基地多输入多输出(multiple input multiple output,MIMO)雷达相干多目标角度快速估计算法。	Put forward a kind of double base multiple input multiple output (multiple input multiple output, MIMO) radar coherent rapid multi-objective Angle estimation algorithm.
55714	采用单次快拍数据构建的一组Toeplitz矩阵重构出新的协方差矩阵,使得矩阵的秩等于目标的总个数;通过矩阵变换使得变换后的协方差矩阵满足centro-Hermitian性,即可利用酉变换将复矩阵转化为实矩阵进行实数域的参数求解,采用旋转不变信号参数估计(estimation of signal parameters via rotational invariance technique,ESPRIT)算法即可准确估计出对应的目标角度。	Using single snapshot data to construct a set of Toeplitz matrix remodeling a new covariance matrix, the matrix rank is equal to the target of the total number of;Through the matrix transformation after the transformation of covariance matrix satisfies, centro - Hermitian complex by unitary transformation matrix can be converted to solving matrix in real number field parameters, adopting rotation invariant signal parameter estimation (estimation of signal parameters via rotational invariance technique, ESPRIT) algorithm can accurately estimate the corresponding target Angle.
55715	通过仿真表明:所提算法能够实现相干目标源的角度估计且参数自动配对;降低了复矩阵特征分解所带来的巨大运算复杂度,具有更高的参数估计精度和算法稳健性;同时能够对移动目标进行角度跟踪,具有良好的跟踪性能。	By simulation show that the proposed algorithm can achieve coherent source Angle estimation and parameters for automatic matching;Reduce the decomposition of complex characteristics caused by the huge computation complexity, parameter estimation precision and the algorithm has higher robustness;At the same time can carry on the Angle tracking of moving target, has the good tracking performance.
55716	卫星导航系统接收机小型化是未来发展的趋势,基于小型导航接收机的无线传感器网络具有新优点的同时也面临新的问题。	Satellite navigation receiver miniaturization is the development trend of the future, wireless sensor network based on small navigation receiver has the advantages of the new also faces new problems.