查询词典 NN

As a result, for feed-forward NN model this paper first proposes the concepts of"attractor"and"attraction region","fake attractor"and"fake attraction region", and norm of describing"attraction region". Further this paper presents a mathematics method of calculating feed-forward NN"attraction region", it solves the overelaborate job of checking NN fault-tolerance performance by testing all detecting samples, so to test fault-tolerance performance using theoretic characteristic norm is an important break. In accordance with feed-forward NN low fault-tolerance and difficult to satisfy practical application, according to practical application theory, this paper presents an actual method of enhancing NN fault-tolerance performance, builds the mechanism of reconstructing NN model based on combining NN and fuzzy process, reforms NN "fake attractor"and"fake attraction region"to enhance NN fault-tolerance performance greatly.

为此，本文首先给出了前馈NN模型的&吸引子&和&吸引域&，&伪吸引子&和&伪吸引域&的定义，以及描述吸引域的指标，在此基础上提出了一种计算前馈NN模型&吸引域&大小的数学计算方法，该方法提出解决了目前依靠对检测样本的测试工作来检验NN容错性的繁琐工作，而用具理论意义的性能指标来检验，这是一个重要的突破；针对前馈NN容错性较低以致难于满足实用的缺点，根据NN实用稳定性理论，提出了一种提高其容错性的实用方法，建立了基于NN与模糊处理相结合的NN模型重构机制，重构NN&伪吸引子&和&伪吸引域&来大大地提高NN的容错性。

the manual nn as a highly integrated chiasma subject. most of models about nn are adopting the bp network and the changed form at the practical appliance, which is also the hard core of forward network, incarnating the essential part of nn.

人工神经网络作为一门高度综合的交叉学科，在实际应用中绝大部分的神经网络模型是采用bp网络和它的变化形式，它也是前向网络的核心部分，体现了人工神经网络最精华的部分。

Moreover, the radius of attraction of memorized sample patterns is presented. A Hybrid neural network model for pattern recognition is presented. The model consists of a cascade connection of three neural networks, which are hetero—associative NN、2—D local—connected NN and classification NN respectively.

提出了一种适于模式识别的复合神经网络，它由三个小网：异联想记忆网、局域连接神经网和分类网级联而成，其输入为模式的二维自相关系数，输出为预定的编码，此网可以有效地进行平移不变模式识别，且有很好的容错性和抗噪声能力。

A control strategy using Mutual Mapping Neural Network to compute the solution of the backward and forward kinematics was presented in this paper. MMNN was consisted of two NN and one modify function, one NN was used for forward kinematics calculation, and another for Jacobi matrix calculation, the modify function was formed based on Lyapunov function.

双向映射神经元网络主要由两个神经网络和一个修正函数构成，其中一个神经网络用来进行正运动学求解，另一个神经网络用来计算雅可比矩阵，修正算法是根据李雅普诺夫函数来计算的。

It is defined that the two integrating ways' characteristics of the wavelet NN and theirs applying on the equipments' intelligence diagnose. Wavelet analysis technology has the ability of analyzing strongly the non-placidity signals' time-frequency, self-studying the NN, of dealing with data distributed and collaterally, of mapping non-linear ability, and it is fit for diagnosing failures.

利用小波分析对非平稳信号强的时频分析能力和神经网络的自学习、分布式并行处理数据及非线性映射能力，以及它们适合于故障诊断的特点，说明小波与神经网络的松散型和和紧致型结合途径下产生的两种小波神经网络的各自特点及其在设备智能诊断中的应用。

The cause of the target fade-out in Neural Network nonuniformity correction scheme for infrared focal plane arraywas studied. Based on analyzing the strength and limitation of edge-directed NN scheme and nonuniformity correction combining one-point calibration and NN-NUC, a new combinational algorithm for nonuniformity correction was proposed.

中文摘要：分析了传统神经网络非均匀性校正算法在空域处理过程中产生目标退化的原因，在总结基于边缘指导的神经网络校正算法与一点定标和神经网络结合的校正算法的基础上，提出了新的组合校正算法。

The project combine the NN model consult self adapt modern intelligent control theory and current digital technology and apply this theory to follow adjust the moment of electromotor, which correspond control the moment of many electromotor. Inc a mainframe (adopt speed and current both closed loop system), seven frae electromotor (adopt NN model consult self adapt adjuster).

将神经网络模型参考自适应现代智能控制理论与当前的数字技术相结合应用于电机的力矩跟随调节，协调控制多台电机的力矩，包括：一台主机(采用速度、电流双闭环系统)，七台从机(采用神经网络模型参考自适应调节器)。

By studying and analyzing the various Neural Network and intelligent fault diagnosis, associating the difficulty which is to measure the interior signal of cycloconverter, the thesis puts forward the cycloconverter fault diagnosis which bases on Neural Network and takes cycloconverter input voltage and output current as the input of NN. According to the self-learning and non-linearity of NN, it can confirm the position of SCR fault.

通过对国内外多种神经网络和智能故障诊断技术的分析比较，并考虑到直接检测变频器内部有关信息的难度，本文提出并研究了基于神经网络的变频器故障诊断技术，以变频器输入/输出有关电量特征值作为神经网络的输入，通过神经网络的自学习方式，对变频器内部晶闸管故障进行定位。

Consequently the NN adaptive unit is added into the control loop according to the NN on-line approximate property. Lastly the chapter combines the proposed shortest path-planning method and the adaptive tracking method in order to stabilize the NWMR about a posture through the shortest path. Many simulations are carried out on MATLAB and the simulation results are very feasible and reasonable. Chapter 5 proposes a new localization method for the WMR running on inclined planes based on fusion of clinometric and an odometric data by Kalman filter.

本文首先求得了在斜面上运行的NWMR动力学系统的状态方程；接着选择了适当形式的系统输出方程，采用非线性系统解耦方法对该系统实现了输入—输出线性化，并应用线性系统方法进行镇定；然后提出了基于人工神经网在线逼进NWMR动力学特性的自适应跟踪控制方法；最后将本文最短距离路径规划方法与跟踪控制方法相结合实现了NWMR系统状态镇定，并通过仿真实验证明了该方法的有效性。

Spd and Spm evidentlydropped more rapidly in the abortive ovules than those in the normal ones,which indicats that lower contents and more rapid decrease of Spd and Spmduring the embryogenetic process might have a close relationship withembryo abortion.The studies also showed that iow ratios of nn-1and nn-1might hinder embryo development.

败育胚珠中的Spd和Spm含量在胚胎败育期的下降速度显著大于正常胚珠，两者含量低以及在胚珠发育进程中较大的降幅与胚胎败育密切相关。nn-1和nn-1比值低亦不利于胚胎发育。

In the days of sailing vessels the crew were afraid they would be becalmed on the ocean.

在使用帆船的时代，船员们担心他们会因为无风而无法在海洋上航行。

As long as foreign donors pay the PA's salary bill, few expect a new intifada.

只要外国继续为巴权力机构的薪水买单，希望发动新暴动的人便寥寥无几。