grnn神经网络

A. GRNN（广义回归神经网络）和BP神经网络

一般用SOM网络 推荐你一本书 神经网络43案例 里头有数字识别这个

B. matlab grnn预测神经网络怎么用

轮台歌奉送封大夫出师西征(岑参)[4]

C. elman神经网络能够解决的问题,还有其他什么网络能够更好的解决

还可以使用GRNN神经网络，效果非常好，并且训练速度非常快。广义回归神经网络GRNN：径向基神经内元容和线性神经元可以建立广义回归神经网络，它是径RBF网络的一种变化形式，经常用于函数逼近。在某些方面比RBF网络更具优势。
在MATLAB中，直接使用net=newgrnn(P,T,spread)就能以非常快的速度设计出一个GRNN网络，其进行训练及预测时，效果非常好，不会比elman神经网络差。扩展常数SPREAD不能太小，才能使部分径向基神经元能够对输入向量所覆盖的区间产生相应，但也不能太大，否则计算困难。可以通过试凑来获得最佳扩展常数。

D. 请教RBF神经网络高手。用matlab设计newgrnn广义回归神经网络，进行训练、仿真、拟合。画出预测图和误差图

这个拟合了也没有多大的意义。
一、数据太少。二、发病率和时间存不存在因果关系还是个疑问

E. 帮忙翻译一下啊！谢谢了！基于GRNN神经网络的中厚板轧机宽展预测模型

Plate Plate Mill is the main rolling equipment, shoulders rolling steel pressure vessels, Offshore Steel, ships and other high-quality steel-plate proction tasks. In order to improve the proction process of rolling automation level and proction efficiency, need for the rolling process of plate deformation accurate prediction and control. Spread change prediction model is rolling width control technology set the width of the core functions of calculation, its accuracy will have a direct impact on the width of the finished proct control effect. Only by correctly calculated and the estimated spread of size, can be rolled out in line with the requirements of the procts. According to the same size and principles, spread by the size of the volume can be calculated directly for the blank size, So research process of rolling wide changes in the size of the development is of great significance. Is necessary considering the plate mill modern proction methods of operation and modern art, Forecast to achieve automatic control system SL process of deformation spread. Spread a theoretical analysis and development of wide deformation refers to the rolling workpiece width along the direction of deformation. Rolling in different conditions, the blank in the process of rolling spread along the height of the cross-sectional analysis shown in Figure 1. Spread △ Says generally following components : sliding spread △ Yue. The figure △-wide outreach and drum-wide outreach △ B3 [11 o spread is a complex deformation of the deformation process. in the process of rolling wide impact on the development of many factors. Spread with a series of rolling factors have complicated relations : △ B = f (H, h, l, B, D, ψa, Δh. ε, f, t, m, Pδ, V, ε) (I) where : a ψ, H, h for the SL system before and after the deformation zone thickness; I, B, D deformation zone for the length, width and roll diameter; deformation of the cross-sectional shape; Δ h, ε, to pass, Rection; . F, t, m coefficient of friction, rolling temperature, the chemical composition of metal; P δ for the metal mechanical properties; V, ε roller linear velocity and deformation rate. From the above analysis shows that the sheet rolling process, not only with the width of the thickness rection has increased, sheet width but also by the process of rolling many factors, these factors between the relations are very complicated, difficult to traditional mathematical models to achieve changes in the spread forecast precision. If considering various parameters on the workpiece width of change, using a mathematical model expression is extremely complicated, can not be directly applied to engineering practice. For 42 oo ~ L machine width, the mathematical model for the information have also introced, Researchers at home and abroad to establish the spread mathematical model has a very wide range, but mainly on the basis of the measured data of the scene, Spread consider the influence of changes in several key factors, ignore other factors, the use of regression analysis of the mathematical model, This is bound to cause certain error; and the spread of existing model are limited in their scope of application. Neural network is highly nonlinear processing capability, operation and high accuracy on-line real-time response characteristics, In describing the spread on the deformation has obvious advantages. Therefore, the deformation of the spread of automatic detection and forecast process, the use of artificial neural network, introction GRNN (GRNN) model algorithm to achieve the width deformation forecast, Rolling to the actual proction process spread deformation control provide precise control basis.

F. 在RBF神经网络预测中，如果样本数据较少时，效果怎么样和GRNN神经网络比较

效果都不好,样本少最好是用统计学加一些机械学习的思想自己多做尝试,设计回特征给出一个答固定的预测模型,效果根神经网络不在一个档次上,人的智商远远比电脑高,神经网络勉强算大数据的技术,前提就是要有大量冗余的数据,大到疲于用统计学方法处理.

G. 急需一个GRNN神经网络的C的源代码，有谁能帮我写一下。实在是看不懂。。。

抱歉！这个实在不容易，没有

H. 通用回归神经网络是啥类似BP神经网络嘛

一般的数据拟合，传统的lsqcurvefit和lsqnonlin，如果较新的方法就很多了，比如神经网络，小样本的一般是GRNN和灰色神经网络，大样本下更多选择了，BP、SVM等等，还可以有遗传算法等等

I. matlab神经网络43个案例分析 百度云搜索

共有43章，内容涵盖常见的神经网络(BP、RBF、SOM、Hopfield、Elman、LVQ、Kohonen、GRNN、NARX等)以及相关智能算法(SVM、决策树、随机森林、极限学习专机等)。同时，部分章节也涉及了常见的优化算法(遗传算法、蚁群算法等)与神经网络的结合问题。此外，本书还介绍了MATLABR2012b中神经网络工具箱的新增功能与特性，如神经网络并行计算、定制神经网络、属神经网络高效编程等。