Integratedneuralnetworkbasedtestfile

PWMSound.m processing signal digital

说明：  本源码是基于遗传算法的多维函数值的优化计算，利用matlab仿真(The source is based on genetic algorithm optimization of multi-dimensional function value calculation, the use of matlab simulation)

利用FLICM算法进行图像聚类的matlab程序代码，相对于FCM（模糊C均值聚类）算法具有较强的抗噪能力。(FLICM image clustering algorithm matlab code, with respect to FCM algorithm (Fuzzy C-means Clustering) it has a strong anti-noise ability.)

neural network trained with unscented kalman filter

二维码制作工具，提供了QR二维码的制作，可以提供大小的标识。(Two dimensional code production tools, provide production QR two-dimensional code, can provide the size of the logo.)

SAX比其他符号算法更简便、高效 在符号化过程中实现了减维降噪 ,保证在符号空间计算出的两个符号序列距离满足实际两个时间序列距离的下界要求 ,即不会出现漏报(SAX (Symbolic Aggregate approXimation))

一般情况下，我们都是在MATLAB命令行或DOS命令行下编译MEX程序。 所用的命令就是：mex filename.c 这有很多不方便的地方： a. 虽然mex也可以编译C++的mex程序，但是它的主框架仍是C的 a. 当程序有多个模块时，需要多次使用mex命令，操作很麻烦 b. 不能利用VC特有的ClassWizard自动创建和维护类 c. 不能用MFC类库编写应用程序 d. 不能方便地进行类似VC的项目管理 本文详细解说如何在IDE中编译MEX程序(Under normal circumstances, we are all in the MATLAB command line or DOS command line program to compile MEX. Command is used by: mex filename.c This has a lot of inconvenient places: a. Although the mex can also compile C++ The mex program, but it is still the main frame C of a. When the procedure has a number of modules the need to repeatedly use the mex command, the operation is cumbersome b. Can not use VC-specific ClassWizard automatically create and maintain the category c. can not use MFC class library to write applications easily d. should not conduct a similar project management VC a detailed explanation of how this article IDE compiling MEX procedures)

IIR滤波器设计，分离出250hz,500hz,1000hz单频(IIR filter design, isolated 250hz, 500hz, 1000hz single frequency)

部分实现了追踪测速迭代松弛算法，matlab开发工具箱中的支持向量机，主要是基于mtlab的程序，通过反复训练模板能有较高的识别率，利用贝叶斯原理估计混合logit模型的参数，BP神经网络用于函数拟合与模式识别，FIR 底通和带通滤波器和IIR 底通和带通滤波器。( Partially achieved tracking speed iterative relaxation algorithm, matlab development toolbox support vector machine, Mainly based on the mtlab procedures, Through repeated training jSOURGDlate have higher recognition rate, Bayesian parameter estimation principle mixed logit model, BP neural network function fitting and pattern recognition, Bottom-pass and band-pass FIR and IIR filter bottom pass and band-pass filter.)

估算带宽，先是计算输入信号单边带功率谱密度，然后估算出信号所占的带宽(Bandwidth estimation, first calculating the input signal is single sideband power spectral density, and then estimate the proportion of signal bandwidth)