ID 原文 译文
50107 为解决当飞行器出现破损以及飞行器受干扰时的轨迹优化问题，新一代的再入式飞行器需要有实时在线轨迹优化的能力。 Neural network dynamic optimization algorithm (neural dynamic optimization, NDO) main characteristic is to make neural network approximation optimal solution.
50108 神经网络动态优化算法可以避免传统的间接法在求解轨迹优化问题时协态变量初值猜测问题。 Neural network dynamic optimization algorithm can avoid the traditional indirect method in solving trajectory optimization problem of state variable initial guess.
50109 仿真结果表明神经网络动态优化算法可以很好的避免协态变量初值猜测问题，具有较强的鲁棒性，能满足实时性要求。 Simulation results show that neural network dynamic optimization algorithm can be very good avoid association, state variable initial guess problem, has strong robustness and can meet the real-time requirements.
50110 给出了神经网络动态优化的原理，详细介绍了优化流程。 The principle of neural network dynamic optimization is given, and the optimization process are also described in detail.
50111 针对属性权重和阶段权重完全未知且属性信息表示为混合形式的多阶段决策问题，提出一种新的决策方法。 In view of the attribute weights and phase weights are unknown completely and the attribute information is expressed as hybrid forms of multistage decision problem, put forward a new method for decision making.
50112 首先，将混合型属性信息统一成区间数形式，在此基础上根据属性信息的熵权区间和离差水平确定属性权重。 First, mixed attribute information unified into the interval form, on this basis, according to the attribute information entropy interval and dispersion level to determine the attribute weights.
50113 然后，借助有序聚类法将决策对象各个阶段的矢量信息划分为若干个聚集，再以聚集中决策矢量的距离最小化为目标，构建优化模型求得聚集内各矢量的阶段权重，进而得到所有决策对象的综合阶段权重。 Then, using the orderly cluster method to decision-making object vector information is divided into several stages of gathered themselves together, and then to gather in the decision vector minimizing the distance of the target, build the optimize model obtained gathered within the weight vector of the stage, the synthesis weights all decision-making object is obtained.
50114 最后，利用TOPSIS法对决策对象进行排序，并通过算例对该方法的可行性和实用性进行证明。 Finally, using the method of TOPSIS to sort by decision-making object, and through the example to prove the feasibility and practicability of this method.
50115 提出一种基于差准则的二维非参数特征分析(2-dimensional nonparametric feature analysis based on difference criterion，2DDNFA)的图像特征提取方法，它结合了二维线性判决分析(2-dimensional linear discriminant analysis，2DLDA)、最大散度差(maximum scatter difference，MSD)、非参数判决分析(nonparametric feature analysis，NFA)3种方法的思想。 In this paper, a two-dimensional nonparametric feature analysis based on differential criterion (2 - dimensional nonparametric feature analysis -based on difference criterion, 2 ddnfa) image feature extraction method, it is a combination of two dimensional linear decision analysis (2 - dimensional linear discriminant analysis, 2 dlda), maximum differential divergence (maximum scatter difference, MSD), nonparametric decision analysis (nonparametric feature analysis, NFA) the ideas of the three methods.
50116 首先利用二维图像样本的近邻样本构造类内、类间散布矩阵，再基于差准则计算投影矩阵，最后将二维图像向投影矩阵投影得到特征矩阵。 Using the data of two-dimensional images near to the first sample scattering matrix between structure within the class, class, and then based on differential criterion to calculate the projection matrix, finally the two-dimensional image to get the characteristic matrix projection matrix projection.