LS-SVM

说明：  这个功能是用于Matlab中的某个功能所使用的，非常好的一个原代码！(This feature is used in a Matlab function used, very good a the original code!)

composit efunction with matlab

基于matlab的各种EOF分解程序，包括旋转EOF分解，复数EOF分解等，包中带详细的例子(Matlab based on EOF decomposition of the various procedures, including rotary EOF decomposition, complex EOF decomposition and so on, the package with detailed examples)

维纳滤波的MATLAB环境实现，使用LMS算法，可用于滤波，图像恢复等应用(Wiener filtering MATLAB environments, the use of LMS algorithm, can be used to filter, image restoration applications)

本设计主要对4PSK调制方式的信号，利用MATLAB的m文件进行最佳接收机的设计与仿真。对输入的叠加噪声的4PSK调制信号进行接收，利用相关解调器来实现信号解调，及最大似然准则来实现检测器。在相关解调器中，接收信号分别与基函数 和 相乘再积分。在检测器中，利用相位来判断输出，从而最终得到接收的数据。采用随机二进制数通过4PSK调制后叠加高斯白噪声再对设计的接收机进行测试，从测试的结果可看出，在信噪比大于-8dB时，误码率为0，说明该接收机较好的实现了抗噪声性能。(The design of the main 4PSK modulation mode of the signal, the use of MATLAB m file for the best receiver design and simulation. The 4PSK modulated signal of the input superimposed noise is received, and the demodulator is realized by the relevant demodulator, and the maximum likelihood criterion is used to realize the detector. In the associated demodulator, the received signal is multiplied by the basis function and multiplied by the basis function. In the detector, the output is judged by the phase, and finally the received data is obtained. Using the random binary number through 4PSK modulation after the superposition of white noise and then the design of the receiver to test the test results can be seen in the signal to noise ratio greater than-8dB, the bit error rate is 0, indicating that the receiver is better The realization of the anti-noise performance.)

说明：  短期电力负荷预测，支持向量机结合粒子群算法进行短期电力负荷预测的实例(short-time load forecasting)

在相空间重构中，用互信息法求最小关联嵌入维(In phase space reconstruction, the use of mutual information method associated minimum embedding dimension)

fastICA2.5 ICA算法，使用快速独立分量分析（ICA）算法进行特征提取(fastICA2.5 ICA algorithm, the use of fast independent component analysis (ICA) feature extraction algorithm)

对信号进行稀疏分解并投影，然后对信号进行重建(Sparse decomposition of the signal and projection, and then reconstruction of the signal)

设r是f(x) = 0的根，选取x0作为r初始近似值，过点（x0,f(x0)）做曲线y = f(x)的切线L，L的方程为y = f(x0)+f (x0)(x-x0)，求出L与x轴交点的横坐标 x1 = x0-f(x0)/f (x0)，称x1为r的一次近似值。过点（x1,f(x1)）做曲线y = f(x)的切线，并求该切线与x轴交点的横坐标 x2 = x1-f(x1)/f (x1)，称x2为r的二次近似值。重复以上过程，得r的近似值序列，其中x(n+1)=x(n)－f(x(n))/f (x(n))，称为r的n+1次近似值(Let r is f (x) = 0 root, select the initial approximation x0 as the r, over point (x0, f (x0)) to do the curve y = f (x) the tangent L, L the equation y = f ( x0)+ f ' (x0) (x-x0), find the intersection of L and the x-axis of abscissa x1 = x0-f (x0)/f' (x0), x1 is called an approximation r. Through points (x1, f (x1)) to do the curve y = f (x) the tangent, and find the intersection of the tangent with the x-axis of abscissa x2 = x1-f (x1)/f ' (x1), x2 is called r the second approximation. Repeat the process, get an approximation of the sequence r, where x (n+1) = x (n)-f (x (n))/f ' (x (n)), as an approximation of r n+1 times)