极小值

According to the characteristics of the gray projection curves, a pair of maximum and minimum points are found near the central position of each gray projection curve. The differences in position are calculated respectively and the scale change parameter is obtained as the ratio of both differences.

此时，可根据尺度变化下图像灰度累积投影曲线的特点，分别计算参考帧和当前帧的投影曲线上一对极大—极小值点的坐标差值，以它们的比值作为缩放系数，实现实时求取。

From the strength of the cylinder in the platen press ,the relations between the allowable stress of material to make the cylinder and the hydraulic pressure are derived when the outer diameter of the cylinder is the minimal value.

根据平板硫化机液压缸圆筒部分的强度条件，推导出液压缸外径为极小值时，液压缸材料的许用应力与工作液压力之间的关系，从而为平板硫化机的优化设计提供了最佳工作液压力值。

Fuzzy h-prototypes algorithm is one of the most effective algorithms of cluster analysis,however,the problem of sensitive to initial value and vulnerable to the problem of local minimum exists.

模糊h-prototypes算法是当前聚类分析中最有效算法之一，但是存在对初始值敏感、容易陷入局部极小值的问题。

If it is merely less than at any nearby point, it is a relative, or local, minimum.

如果仅仅小于附近任一点的值，那它就是相对的，或者说是局部的极小值。

An extremely high temperature of anode humidification is also unreasonable; as a result, when the current density is high, the liquid water should be injected directly into anode gas channel to guarantee both high inlet H〓 concentration and well-humidified membrane. 4. The utilization of Pt in the cathode catalyst layer will decrease with the increasing current density; as a result, based on the principle of maximal Pt utilization, the catalyst layer can be designed much thinner or not uniformly in the catalyst distribution.

而在此截面上，交指型流场燃料电池阳极侧的液态水饱和度的极值位置并不确定；在电池阴极处，液态水饱和度的极值总是出现在z=0.002m处的气体扩散层和膜的界面上，并且该处的液态水饱和度沿x方向存在一个最大值，与此相应的氧气的质量浓度分数在该处达到极小值，因此极限电流密度将首相出现该处，但总体来说液态水饱和度沿x方向的分布较为均匀。4。

There is mutual contradiction between direction estimation of user's signal and power allocation among all users in the beamforming algorithm. A twice game beamforming algorithm based on game theory is proposed to deal with it. Beamforming game algorithm mathematics model is constructed. During the first game, direction of signal and power allocation are mapped the game theory as "player", which are modeled as the problem of maximin function and obtain direction estimation first. Then during the second game, power allocations of different users are described as a multi-user game. Power updated algorithm is designed. The existence and uniqueness of the Nash equilibrium in the twice game beamforming algorithm based on game theory are proved by mathematics derivation.

摘要针对波束成形算法中，用户的信号方向估计值和用户之间的功率分配存在着相互矛盾，本文提出了一种基于博弈论的二次博弈波束成形算法，构建了波束成形博弈算法数学模型，首先在第一次博弈的时候，将波束成形算法中的信号方向和功率分配映射为博弈论数学模型中的"局中人"，将其建模为函数的极大极小值求解问题，先求解出信号方向；然后在第二次博弈的时候，将不同用户的功率分配过程描述为一个多用户的博弈过程，设计了功率分配更新算法，通过数学推导论证了纳什平衡点的存在性和唯一性。

Experimental results shown that the proposed algorithm not only avoids the local optima and is robust to initialization, but also increases the convergence speed.

实验结果表明，该算法不仅有效克服了FCM算法易陷入局部极小值的缺点，而且明显地避免了对初始化选值敏感性的问题，也有较快的收敛速度。

The differences in position are calculated respectively and the scale change parameter is obtained as the ratio of both differences.

此时，可根据尺度变化下图像灰度累积投影曲线的特点，分别计算参考帧和当前帧的投影曲线上一对极大—极小值点的坐标差值，以它们的比值作为缩放系数，实现实时求取。

We also developed several experiential formula based on structural unit model for high-index surfaces, which can give an estimate of the energies of the high-index surfaces if the value of surface energy of (001),(110) or (111) are known.

作为各条曲线表面结构单元的两个低密勒指数面[001]晶带的(100)面和(110)面，[-110]晶带的(110)面和(111)面或(112)面其表面能值分别是对应曲线上的一个极小值点。

Focusing on the disadvantages in the BP neural network such as low convergence rate,easily falling into local minimum point and weak global search capability,a new method that used GA to train BP neural networks to gain best weights was presented.

针对传统的BP算法存在收敛速度比较慢、容易陷入局部极小值以及全局搜索能力弱等缺陷，提出了神经网络与遗传算法。结合的方法，采用遗传算法。练BP神经网络，优化网络权值。

I didn't watch TV last night, because it .

昨晚我没有看电视，因为电视机坏了。

Since this year, in a lot of villages of Beijing, TV of elevator liquid crystal was removed.

今年以来，在北京的很多小区里，电梯液晶电视被撤了下来。