最小均方濾波器是一類可通過最小化誤差訊號之均方值而修正濾波器係數，以類比所需理想濾波器的自適應濾波器，其中作為修正依據的誤差訊號為理想參考訊號與實際輸出訊號之差。該種濾波器所用之最小均方法只以當前之訊號誤差值為準進行修正，是一種隨機梯度下降法。最小均方法係由斯坦福大學的Bernard Widrow教授及他的首位博士班學生Marcian Hoff於1960年提出。最小均方濾波器屬於常見的自適應濾波器種類之一。
The minimum mean square filter is an adaptive filter that can correct the filter coefficients by minimizing the mean square value of the error signal to analogize the desired ideal filter. The error signal used as the correction reference is the ideal reference signal and the actual output. The difference between the signals. The minimum averaging method used in this kind of filter is only corrected based on the current signal error value, which is a stochastic gradient descent method. The minimum mean method was proposed in 1960 by Professor Bernard Widrow of Stanford University and his first PhD student, Marcian Hoff. The least mean square filter is one of the common types of adaptive filters. (2019-02-21, matlab, 108KB, 下载5次)
LMS算法程序,最小均方算法,感知器和自适应线性元件在历史上几乎是同时提出的，并且两者在对权值的调整的算法非常相似。它们都是基于纠错学习规则的学习算法。感知器算法存在如下问题:不能推广到一般的前向网络中;函数不是线性可分时，得不出任何结果。而由美国斯坦福大学的Widrow和Hopf在研究自适应理论时提出的LMS算法，由于其容易实现而很快得到了广泛应用，成为自适应滤波的标准算法。
LMS,The least mean square algorithm,sensor and adaptive linear elements are proposed at almost the same time in history,and the two algorithms are very similar in the adjustment of weight values.They are all learning algorithms based on error correction learning rules. The sensor algorithm has the following problems: It can not be generalized to the general forward network; When the function is not linear separability, no result can be obtained. The LMS algorithm proposed by Widrow and Hopf of Stanford University in the United States when studying adaptive theory was quickly widely used because of its easy implementation and became a standard algorithm for adaptive filtering. (2018-09-27, C/C++, 1KB, 下载1次)
BFGS算法(BFGS algorithm)，是一种逆秩2拟牛顿法。Hk+，满足拟牛顿方程的逆形式Hk+}少一、k=s.当Hk正定且(，'，少)}0时Hkh，也正定，因此，由BFGS修正确定的算法xk+} - xk - HkF Cxk)是具有正定性、传递性的拟牛顿法，它也是无约束优化中最常用的、最稳定的算法之一这种算法是布罗依丹(Broy-den,C. G.)于1969年，以及弗莱彻(Fletcher , R. ) ,戈德福布(Goldforb,D. )、香诺(Shan no, D. F.)于1970年分别得到的，故称为BFGS算法.
The BFGS algorithm is an inverse rank 2 quasi-Newton method.Hk +, to satisfy the inverse form of the quasi-Newton equation Hk +} less one, k = s. When Hk is definite and (, ', less)} 0 Hkh, is also positive definite, therefore, the algorithm determined by the BFGS correction xk +} - xk - HkF Cxk) is a quasi-Newton method with positive definiteness and transitivity. It is also one of the most commonly used and most stable algorithms in unconstrained optimization. This algorithm was Broy-den (CG) in 1969, and Fletcher (R.), Goldforb (D.), and Shanno (DF) were obtained in 1970 respectively, hence the BFGS algorithm. (2018-03-17, WINDOWS, 1KB, 下载3次)
2014年全国数学建模折叠圆桌模型程序源码
一种桌面呈圆形，桌腿随着铰链的活动可以平摊成一张平板的可折叠桌子桌腿由两组各用一根钢筋连接的木条组成，钢筋两端分别固定在桌腿各组最外侧的两根木条上，并且沿木条有空槽以保证滑动的自由度。
问题一：给定长方形平板尺寸为120 cm × 50 cm × 3 cm，每根木条宽2.5 cm，连接桌腿木条的钢筋固定在桌腿最外侧木条的中心位置，折叠后桌子的高度为53 cm。试建立模型描述此折叠桌的动态变化过程，在此基础上给出此折叠桌的设计加工参数
2014 National Mathematical Modeling Folding Round Table Model Open SourceOne kind of circular table, table legs with hinges activities can be amortized into a flat folding table and legs by the two groups with a reinforced connecting strips of wood, steel ends are fixed to the legs of each group the outermost two strips of wood, strips of wood along the grooves and free to slide in order to ensure freedom.One problem: given a rectangular plate size 120 cm × 50 cm × 3 cm, each wood wide 2.5 cm, reinforced wooden legs connected fixed to the center position of the outermost legs of wood, the table is folded height 53 cm. Try to establish this model to describe the dynamic process of folding tables, folding tables are given on the basis of this design and processing parameters (2016-07-13, PPT, 2KB, 下载7次)
设计要求：试设计一个演示程序（可以采用任何语言编写，建议使GUI工具构建演示程序），该程序运行后能显示一个人机交互界面，该界面允许输入系统状态空间模型的参数矩阵（即可输入A,B,C,D矩阵，以构建系统的状态空间模型），根据输入的参数矩阵，能够给出系统是否能控的结论。若系统能控，则允许输入任意初始状态向量和控制作用的持续时间（即给定系统状态达到零状态的时间长度），能自动构建控制作用，使系统的状态变成零状态，系统状态从给定状态变成零状态的过程采用图形的方式展示。
Design requirements: try to design a demonstration program (written in any language can be used, it is recommended to make GUI tool for building demonstration program), after running the program can display a human-machine interface, the interface allows parameter matrix input state space model (to enter the a, B, C, D matrix, to construct the system state space model), based on the input parameter matrix, can give conclusions whether the system can be controlled. If the system can control, then allows you to enter the duration of an arbitrary initial state vector and control action (ie a given system state reaches zero the length of time), can automatically build control effect, so that the system state to the zero state, the system state given state to the zero state process using graphically show. (2016-06-13, Visual C++, 11KB, 下载6次)
一CQMFB，低通滤波器H0 (z)来自一半带滤波器。该半带滤波器的长度为47，通带截至频率ω_p = 0.42π，试给出H0 (z)，H1(z)，G0(z)，G1(z)的幅频响应，单位抽样响应并画图。试着产生一信号，它由两个正弦加白噪声组成，一个在低频，一个在高频，正弦的频率及和白噪声的信噪比自己给定。试用所设计的滤波器组对该信号进行分解和重建。比较重建后的效果，并对结果继续进行分析。
A CQMFB, low-pass filter H0 (z) half band filter.The half band filter length of 47, passband by frequency = 0.42 PI, try to give H0 (z), H1 (z), G0 (z), the amplitude frequency response of G1 (z), unit sample response and drawing.Try to generate a signal, it is composed of two sine added white noise, one is in low frequency, one in high frequency, the frequency of the sinusoidal and white noise and the signal-to-noise ratio of a given himself.Trial designed for the signal decomposition and reconstruction filter group.Comparing the effect of reconstruction, and continue to analysis the results. (2016-03-17, matlab, 1KB, 下载6次)
在一个移动通信系统中，发射信号是一个未调制的音频x(t)=cos(2πft)，其中f=f0+fi。设标称载波为f0=1GHz，并设fi =512Hz。假定移动用户在一个繁忙的高速公路上距离基站l公里处的地方，从开始从每小时60公里的速度开始，其速度（在0.01秒的时间间隔内测量）以1/2的概率随机的增加或减少1/10公里每小时。试仿真移动用户在这2s的时间间隔内所接收到的信号(及其频谱)。接收信号可以表示为：
y(t)=a(t)Cos[2π(fo+f1)(t-△t)]
式中 △(t)=d(t)/c
且 a(t)=k/d2(t)
在上面的表达式中d(t)是在t时刻基站与移动用户的距离，c是光速，△(t)是传播延迟，而a(t)是信号在t时刻的衰减，并假设在这2s内a(t)是一个常数。利用复低通等效模型来进行仿真。
In a mobile communication system, the transmitted signal is an unmodulated audio x (t) = cos (2πft), where f = f0+ fi. Let nominal carrier of f0 = 1GHz, and set fi = 512Hz. Assume that mobile users on a busy motorway station l kilometers place, starting the beginning of 60 kilometers per hour, the speed (measured in 0.01 second interval) to increase the probability of a random 1/2 or reducing 1/10 km/h. Simulation test signal which the mobile subscriber in the 2s time intervals of the received (and its spectrum). The received signal can be expressed as:
y (t) = a (t) Cos [2π (fo+ f1) (t- △ t)]
Where △ (t) = d (t)/c
And a (t) = k/d2 (t)
In the above expression, d (t) at time t is a base station and a mobile user s distance, c is the speed of light, △ (t) is the propagation delay, and a (t) is the attenuation of the signal at time t, and assuming that 2s within a (t) is a constant. Using complex low-pass equivalent model for emulation. (2015-11-24, matlab, 2KB, 下载8次)
.已知某炉温控制系统，要求温度保持在600℃恒定，针对该控制系统有以下控制经验：
（1）若炉温低于600℃，则升压，低的越多升压越高；
（2）若炉温高于600℃，则降压，高得越多降压越低；
（3）若炉温等于600℃，则保持电压不变。
设模糊控制器为一维控制器，输入语言变量为误差，输出为控制电压。输入输出变量的量化等级为7级，取5个模糊集，试设计隶属度函数误差变化划分表，控制电压变化划分表和模糊控制规则表。
Known for a temperature control system that requires a constant temperature is maintained at 600 ℃, the following control experience for the control system:
(1) If the furnace temperature is below 600 ℃, the step-up, the more the higher the lower the boost
(2) if the furnace temperature is higher than 600 ℃, the blood pressure, high blood pressure, the lower the more
(3) If the furnace temperature is equal to 600 ℃, then keep the voltage constant.
Set of fuzzy controller for one-dimensional controller, input linguistic variables for error, the output control voltage. Input and output variables of quantization levels for seven, take five fuzzy sets, membership function test design error variance divided table, partition table and the control voltage variation fuzzy control rule table. (2015-04-10, matlab, 19KB, 下载48次)
设计一CQMFB，低通滤波器H_0 (z)来自一半带滤波器。该半带滤波器的长度为47，通带截止频率ω_p=0.42π，试给出H_0 (z)，H_1 (z)，G_0 (z)，G_1 (z)的幅频响应，单位抽样响应并画图。试着产生一信号，它由两个正弦加白噪声组成，一个在低频，一个在高频，正弦的频率及和白噪声的信噪比自己给定。试用所设计的滤波器组对该信号进行分解和重建。比较重建后的效果，并对结果继续进行分析。
Design a CQMFB, low pass filter H_0 (z) the half-band filter. The length of the half-band filter 47, passband cutoff frequency ω_p = 0.42π, try to give H_0 (z), H_1 (z), G_0 (z), G_1 (z) of the amplitude-frequency response, the unit sample response and Paint. Try to generate a signal that consists of two sinusoidal composition plus white noise, a low frequency, a high frequency in the signal to noise ratio, frequency and sine and white noise of its own given. Trial designed filter group that signal decomposition and reconstruction. Results were rebuilt, and the results continue to be analyzed. (2015-03-23, matlab, 1KB, 下载8次)
图1为均衡带限信号所引起失真的横向或格型自适应均衡器(其中横向FIR系统长M=11), 系统输入是取值为±1的随机序列，其均值为零；参考信号；信道具有脉冲响应：
式中用来控制信道的幅度失真(W = 2～4, 如取W = 2.9,3.1,3.3,3.5等)，且信道受到均值为零、方差(相当于信噪比为30dB)的高斯白噪声的干扰。试比较基于下列几种算法的自适应均衡器在不同信道失真、不同噪声干扰下的收敛情况(对应于每一种情况,在同一坐标下画出其学习曲线):
1）横向/格-梯型结构LMS算法
2）横向/格-梯型结构RLS算法
并分析其结果。
Figure 1 is a band-limited signal caused balanced horizontal or lattice distortion adaptive equalizer (FIR system in which the lateral length M = 11), the system input is the value of a random sequence of ± 1, mean zero reference signal channel has impulse response:
Wherein the amplitude of the distortion for the control channel (W = 2 ~ 4, and so as to take W = 2.9,3.1,3.3,3.5), and subjected to a channel with zero mean and variance (equivalent SNR is 30dB) Gaussian white noise. Compare the following types of algorithm-based adaptive equalizer in a different channel distortion, the convergence of different noise interference (corresponding to each situation, draw their learning curve in the same coordinate):
1) landscape/grid- LMS algorithm ladder structure
2) landscape/grid- RLS algorithm ladder structure
And analyze the results. (2014-09-21, Visual C++, 3KB, 下载12次)
1. 给定模拟信号：
1）选择采样频率Fs = 5000Hz和合适的信号长度，采样得到序列 x1(n)。求并画出x1(n)及其序列傅里叶变换 |X1(ejw)|。
2）选择采样频率Fs = 1000Hz和合适的信号长度，采样得到序列 x2(n)。求并画出x2(n)及其序列傅里叶变换 |X2(ejw)|。
3） 说明|X1(ejw)|与|X2(ejw)|间的区别，为什么？
2. 已知两系统分别用下面差分方程描述：
试分别写出它们的传输函数，并分别打印曲线。说明这两个系统的区别。
1 Given the analog signals: 1) Select the sampling frequency Fs = 5000Hz and a suitable signal length, sampled sequence x1 (n). Seek and draw x1 (n) and sequence Fourier transform | X1 (ejw) |. 2) Select the sampling frequency Fs = 1000Hz and a suitable signal length, sampled sequence x2 (n). Seek and draw x2 (n) and sequence Fourier transform | X2 (ejw) |. 3) Description | X1 (ejw) | and | X2 (ejw) | difference between, why? (2) Given two systems are described by the following differential equation: respectively try to write their transfer functions, and were printed curve. Explain the difference between these two systems. (2014-04-14, matlab, 1936KB, 下载3次)
1、试设计串连校正装置，在同一窗口下分别绘制出校正前后的频域响应曲线，阶跃响应曲线，判断系统校正前后的稳定性；
2、设计串连校正装置，在同一窗口下绘制出校正前后的波特图进行验证，并在同一窗口下绘制校正前后系统的奈奎斯特图；
3、设计串联滞后超前校正装置，绘制出校正前后系统的伯德图；
4、设计串联校正装置使校正后的闭环系统的主导极点；
5、设计串联校正装置，维持原系统的闭环主导极点基本不变；
6、设计状态反馈阵。
And try to design concatenation correction device, were drawn in the same window, the frequency response curve before and after correction, the stability of the step response curve, determine the system before and after correction design concatenation correction device, drawn in the same window Bode plots before and after a correction to verify, before and after correction system Nyquist diagram drawn in the same window 3 design series lag the leading correction device, drawn before and after correction system Bode plot design series correction The closed-loop system in the device so that the corrected the dominant pole design series correction device to maintain the original system closed-loop dominant pole is essentially the same 6, the design state feedback matrix. (2013-04-22, matlab, 3KB, 下载60次)
自己写的matlab上GUI界面，可以读取行情软件导出的txt格式历史数据，并汇出k线和技术指标，可以通过GUI编辑设置、各种指标及其参数。主图附图指标编辑通过编写脚步实现，有例子。
解压后，将work下的dir.mat中的wkdir变量修改成与你放置的目录一致，也可以就将work放置在"d:/matlab/"下.
read2mat.m从文本中读取数据，并存在与文本文件同名的.mat格式的文件中。
运行kfigure,加载.mat格式文件就可以画出k线图和指标图。其他功能很多和股票软件差不多，大家自己试就行了。
Matlab GUI interface to write their own, you can read the historical data of market software exported txt format, and Export k lines and technical indicators, you can edit the settings through the GUI, various indicators and their parameters. The the main figure drawings indicators edit achieve by writing the footsteps examples. After decompression, modify the work under dir.mat wkdir variable and you are placed in the same directory, will work placed in the " d :/ matlab /" . Read2mat.m to read data from a text, and exist with a text file of the same name. mat format file. The run kfigure, loading mat format file can draw k chart and indicators Figure. Other features and stock software almost everyone own try line. (2012-12-28, matlab, 2192KB, 下载125次)
拉格朗日算法已知三组温度与压力数据:(3,53.5) (4,55.3) (5,56.2)求4.15对应的值为55.492375，结果正确。如果将以上数组顺序颠倒，已知压力55.492375求对应的温度，结果为4.185739，可以看到上下两个结果误差较大，就是说不能象这样去颠倒顺序去求值的。但是实际上我们可能经常要求知道压力去求温度，那么遇到这样的情况，只能采用试算法：一种简单的方法就是二分法迭代，书上都有介绍就不多讲了。
Lagrange method已知三组温度与压力数据:(3,53.5) (4,55.3) (5,56.2)求4.15对应的值为55.492375，结果正确。如果将以上数组顺序颠倒，已知压力55.492375求对应的温度，结果为4.185739，可以看到上下两个结果误差较大，就是说不能象这样去颠倒顺序去求值的。但是实际上我们可能经常要求知道压力去求温度，那么遇到这样的情况，只能采用试算法：一种简单的方法就是二分法迭代，书上都有介绍就不多讲了。 (2010-11-09, matlab, 1KB, 下载4次)
1、指标信号，仅当参考；理解理念，才更重要！指标中大阳支撑的理念，请大家适当参考；其余提示信号，均可只当戏玩，不可生搬硬套！
2、【烧香拜佛】指标极其简单，不可单独使用，请结合盘面热点、板块龙头等情况，综合考虑目标的机会。
3、该指标对于进入主升浪、台阶式上涨的目标，最为有效！
下面给出大智慧L2、飞狐、通达信三种常用软件的版本源码，请需要的朋友适当参考。
1, indicator signal, only when the reference understand the concept, it is more important! Indicators support the idea of the sun, please refer to the appropriate the remaining prompts signal, can only play when the play, not rote! 2】 【pray to Buddha indicator is extremely simple, is not alone, please combine disk hot plate leading, etc., considering the target opportunity. 3, the indicator for access to the main upward move, stepped up the goal, the most effective! Here are great wisdom L2, flying fox, accessible version of the letter are three commonly used software source code, make a friend in need the appropriate reference. (2010-04-07, TEXT, 1KB, 下载20次)
对混沌时间序列相空间重构C—C方法的三点不足，提出了一种基于改进的C-c方法的
确定曩优时延与嵌入窗的新算法．在关联积分计算过程中引入了权衡计算精度与速度的可调参数，
合理选择试参数，能在不严重损失估计精度的前提下，大大加快计算速度。在理论分析的基础上，
用所提出的算法对三种混沌序列进行相空间重构，仿真结果表明该算法对最优时延的选择更准确，
对最优嵌入窗的选取更可靠。
Of chaotic time series phase space reconstruction methods of C-C less than three points, a method based on an improved determination of former times Cc excellent window of time delay and embedding the new algorithm. Calculation of correlation integral in the process of introducing a trade-off accuracy and speed of the adjustable parameters, a reasonable choice of test parameters can be estimated without serious loss of accuracy under the premise of the calculation speed greatly accelerated. In the theoretical analysis, based on the algorithm used by the chaotic sequence of three phase-space reconstruction, simulation results show that the algorithm of the optimal choice of a more accurate time delay, the optimal embedding window for the selection of more reliable. (2009-04-22, PDF, 276KB, 下载74次)
一个报童从报刊发行中心订报后零售，每卖一份报纸可赚钱a元；若报纸卖不出去，则退回发行处，每退一份要赔钱b元。每天报童卖出的份数是随机的，但报童可以根据以往卖报情况统计得到每天卖k份报纸的概率密度p(k)。
（1） 求报童每天期望收益达到最大（或损失达到最小）的定报量z。
（2） 改变参数a/b的值，观察订报量的最优值变化，画出变化曲线。
试画出仿真流程图，进行程序实现，并对仿真结果进行分析。
A newsboy subscribe from the press after the retail distribution centers, each sale of a newspaper can make money a million if newspapers can not be sold, then returned to Issue Department, each retire to a loss of a b element. Newsboy sold shares in a day is random, but the newsboy can sell based on past statistics reported to be selling daily newspaper k the probability density p (k). (1) seeking newsboy expected profit per day to maximize (or minimize loss) reported that the volume of fixed-z. (2) change the parameters a/b values, subscribe to observe changes in the volume of the optimal value, draw curve. Simulation test to draw a flow chart for program realization, and simulation results for analysis. (2008-06-14, matlab, 1KB, 下载24次)