基于 rbf 神经网络的时间序列预测

关于matlab的一本书，吕梁山编写的matlab与通信仿真的书籍，写的很好，很有参考价值(A book on matlab, matlab Luliangshan written books and communication simulation, well written, useful reference)

叠前反演的matlab代码，方法采用马尔科夫共轭梯度法，针对一维模型的反演(Pre-stack inversion matlab code, the method uses Markov conjugate gradient method, inversion for the one-dimensional model of)

说明：  考虑色散以及自相位调制对于长距离光脉冲传输的影响(Considering the influence of dispersion and self-phase modulation on long-distance optical pulse transmission)

说明：  实现风速拟合威布尔分布参数估计及ARMA预测，MATLAB可运行。。。。。(Wind speed fitting Weibull distribution parameter estimation)

典型的光伏电池模型，输出受光照、温度的影响。(classical PV model by matlab)

说明：  MATLAB编写的中值滤波函数，可用于一维数据的平滑处理。(The median filter function written in MATLAB can be used for smoothing one-dimensional data.)

异步电机abo仿真模型基于matlab的simulink仿真 源码-Abo simulation model of induction motor based on the simulink simulation matlab source

hausdorff距离的一种改进算法，是我自己编写的，用于提取焦点之后的图像配准(hausdorff distance of an improved algorithm, which I have written for the extraction of the focus after the image registration)

直流电动机的机械特性，很有用的，MATLAB仿真(DC motor mechanical properties, very useful, MATLAB simulation)

说明：  　本文对变步长自适应滤波算法进行了讨论,建立了步长因子μ与误差信号e(n) 之间另一种新的非线性函数关系. 该函数比已有的sigmoid 函数简单,且在误差e(n)接近零处具有缓慢变化的特性,克服了Sigmoid 函数在 自适应稳态阶段步长调整过程中的不足. 由此函数本文得出了另一种新的变步长自适应滤波算法,并且分析了参数α,β的取值原则及对算法收敛性能的影响. 该算法有较好的收敛性能且计算量少. 计算机仿真结果与理论分析相一致,证实了该算法的收敛性能优于已有的算法.(In this paper, variable step adaptive filter algorithm to discuss the establishment of a step factor μ and the error signal e (n) between another new nonlinear function. The function has a sigmoid function than the simple and the error e (n) close to zero Department has a slow change in the characteristics of a Sigmoid function to overcome the steady-state phase in the adaptive step length adjustment during the process. this function, this paper gives another new variable step size adaptive filtering algorithms, and analysis of the parameters α, β the value principles and the convergence performance. The algorithm has better convergence performance easy to calculate. Simulation results and theoretical analysis, confirmed the convergence of the algorithm better performance than existing algorithms.)