网络函数

It is proved that this algorithm can be used to build a B|spline neural network with minimum hidden units to approximate any continuous function defined on compact set to a prescribed accuracy.

对于3层神经网络而言，输入至输出的映射实际上是通过基函数的加权组合来实现的，当基函数取不同时，网络有不同的性质

A fixed point of a discrete-time neural network with generalized input-output function and by using the converse theorem of Lyapunov function we study the uniformly asymptotical stability of equilibrium in this discrete-time neural network with variable weight and give some sufficient conditions that guarantee the stability of it.

本文首先介绍了几类离散神经网络模型的由来及其研究概况，利用Schauder不动点原理证明了一类具有广义输入输出函数的离散神经网络模型平衡点的存在性，利用Lyapunov函数逆定理给出了这类离散神经网络模型在时变权值下的一致渐近稳定性的充分条件。

In this thesis, we firstly introduce several class of discrete-time neural network models and the research progress of the neural networks. By Schauder fixed-point principle we prove the existence of an equilibrium (i.e. a fixed point) of a discrete-time neural network with generalized input-output function and by using the converse theorem of Lyapunov function we study the uniformly asymptotical stability of equilibrium in this discrete-time neural network with variable weight and give some sufficient conditions that guarantee the stability of it.

本文首先介绍了几类离散神经网络模型的由来及其研究概况，利用Schauder不动点原理证明了一类具有广义输入输出函数的离散神经网络模型平衡点的存在性，利用Lyapunov函数逆定理给出了这类离散神经网络模型在时变权值下的一致渐近稳定性的充分条件。

Characteristics of the Internet traffic data flow are studied based on the chaos theory. A phase space that is isometric with the network dynamic system is reconstructed by using the single variable time series of a network flow. Some parameters, such as the correlative dimension and the Lyapunov exponent are calculated,and the chaos characteristic is proved to exist in Internet traffic data flows. A neural network model is constructed based on radial basis function to forecast actual Internet traffic data flow.

摘 要：应用混沌理论，分析了网络流量，用单变量的网络流量时间序列重构与网络动力系统等距同构的相空间，进而计算了实际网络的关维数和Lyapunov指数，并证实了网络流量存在混沌特性；据此建立了基于径向基函数神经网络的模型，并对实际网络数据流进行了预测。

The main works include five parts:(1) A modified K-means algorithm for optimizing the RBPNNs' structures was proposed;(2) The RBPNNs' structure optimization method based on the orthogonal least square algorithm was proposed that can greatly reduce the computation expense for the structure optimization;(3) Using the GA to perform the whole structure optimization of the RBPNNs was proposed in this thesis, which include simultaneous optimization of the hidden centers and the corresponding controlling parameters of the kernel functions.

主要工作体现在如下三个方面：(1)提出一种改进的 K-Means 算法；(2)提出了基于递推正交最小二乘算法的结构优化算法来训练径向基概率神经网络，从而大大减小了用于结构优化的计算开销；(3)提出使用遗传算法来实现径向基概率神经网络的全结构优化，即隐中心矢量和核函数控制参数同时优化，通过新设计的编码方式、新构造的适应度函数，充分发挥了 GA 的全局搜索性能，使得所优化的径向基概率神经网络的结构趋于最简。

In three-valued neural networks model, the energy function of two-valued neural networks model can not be applied, so the energy function need to be find for the three -valued neural networks model, and here two kind model algorithms are compared in this paper.

在三值神经网络模型中，二值模型的能量函数不能适用于三值模型，因此需要求解三值模型的能量函数，在此本文对三值神经网络测试生成算法与二值神经网络测试生成算法进行比较。

The uniform approximation of normal wavelet neural network and the robust analysis of wavelet neural networks of the combination of Sigmoid function are detailedly introduction; Multiple model failure detection based on wavelet neural network is demonstrated detailedly; At last, the failure diagnosis results of aerocraft is present seperately by employing wavelet neural network and BP neural network, and the fault diagnosis of areocraft system by wavelet neural network is achieved.

论文以小波神经网络为研究对象，提出了一类新的加权小波基，分析证明了加权小波基的诸多良好特性；对于常见小波神经网络的一致逼近特性、S型函数组合小波神经网络的鲁棒性分析、多模型小波神经网络的故障检测等问题给出了详细的论证；最后，针对歼击机的常见故障问题，分别给出了应用小波神经网络和BP神经网络的故障诊断结果，实现了小波神经网络对飞机系统的故障诊断。

According to the actual circumstance of performance evaluation for mechanical products, the general neural network is reformed. Therefore, we construct the fuzzy neural network. This paper expresses membership function in the network by trapezium formulae .The indicators are fuzzed using the membership function .

网络中采用梯形函数的四元组形式构造模糊隶属函数，并用隶属函数的端点值和中间值作为样本，用改进的反向传播算法训练网络，得到与待评价产品情况相符合的权重值。

Improving vertical resolution for density curves is by means of restructuring the response from large and small spacing sondes and borehole effect correction factor and coherence factor are used. Form the principle of GR logging, calculate volumes in investigating areas taken by borehole, formation, up and low shoulder beds, along with distance between center of gravity of the four unit bodies and detector. Based on the above, the weighting function of each unit body to GR signal contributions are established, thus realizing processing of GR logs with high resolution. The resistivity logging model uses "resolution matching" method to improve the resolution of the logs. After the process, PNN model is made to recognize the oil-gas-water bed. Compared with BPNN, the net work is featured swiftness, accuracy and ease of practical application.

提高密度测井曲线垂直分辨率是通过对重新组合长、短源距探测器响应的算法来提高补偿密度曲线的垂直分辨率，并加入井眼影响校正因子和相关性因子；自然伽马测井模型从基本原理出发，扣除井的影响，计算井筒、地层及其上下围岩所占探测球域体积及其重心到探测器的距离，以此建立各单元体对伽马测井信号贡献的权函数，通过权函数逐层计算不同厚度地层条件下各的单元体权因子，实现自然伽马曲线高分辨率处理；电阻率校正模型应用&分配率匹配&模型，应用低分辨率曲线提高为高分辨率曲线：曲线校正后，利用常规测井资料并且结合校正后的密度、伽马、电阻率等三个对判断油水层有利的指标，通过概率神经网络模型识别储层流体性质，较以前的模型，不仅极大的提高识别正确率，同时与普通神经网络相比较，提高了网络的运行速度和实用性。

Based on researching into the classification mechanisms of feedforward single-layer radial basis function and linear basis function networks,the author presents the viewpoints that the RBF centers and widths should be determined through a self-learning procedure,that some new kernels naturally come into being according to which class the labeled patterns are misclassified to.

研究了前向单层径基函数网络和前向单层线性基本函数网络的分类机理，提出了RBF的中心和宽度应通过学习自动确定，在学习过程中根据错分样本被错分入的类别自动生成新的核函数这一观点。

Neither the killing of Mr Zarqawi nor any breakthrough on the political front will stop the insurgency and the fratricidal murders in their tracks.

在对危险的南部地区访问时，他斥责什叶派民兵领导人对中央集权的挑衅行为。

In fact,I've got him on the satellite mobile right now.

实际上 我们已接通卫星可视电话了