essay note

Only as a learning record, Gangster skip.

Baseline drift

Bayesian formula

Portal

Sympathetic, parasympathetic (vagal)

Pulse wave

Off-inch ruler

Statistically significant (P value)

The smaller the P value, the higher the statistical significance

0.05≥P> 0.01 was considered statistically significant, and 0.01≥P≥0.001 is considered highly significant
as P = 0.05 prompt associated variable sample 5% may be due to chance result.

Portal

Signal preprocessing which links comprising

Eliminate baseline drift filter

In an example heart Preprocessing:
① ② baseline drift frequency interference ③ EMG interference
portal

Fourier transformation relationship with the frequency of the sampling frequency of the machine

* Represents the actual physical frequency of physical signals AD acquisition frequency, fs is the sampling frequency, can be known by the Nyquist sampling theorem, fs must be ≥ 2 times the highest frequency signal aliasing will not occur, and therefore can be sampled signal fs maximum frequency fs / 2.

· Angular frequency is 2 * pi times the physical frequency, also known as the analog frequency. (Lu Note: Since a signal cycle (e.g., alternating current) is 360 degrees, so the angular frequency is 2pi i.e. how many turn 2pi angular frequency is provided purely for convenience of calculation.)

· Normalized frequency is the result after actual physical frequencies between fs normalized, the maximum signal frequency fs / 2 corresponding to the normalized frequency of 0.5, which is why the normalized frequency fdtool tool matlab is why the maximum the only reason to 0.5.

· Circular frequency is owned by 2 * pi times of a frequency, also known as the digital frequency. That is the normalized angular frequency.

FFT frequency analysis II about the actual physical frequency

Made of n points FFT, represented in the time domain of the original signal takes to do spectral analysis n points, n-point FFT results in n points remained.

In other words, the digital frequency w 2pi divided into n parts, and the entire digital frequency w covering the range from the simulated frequency range of 0-2pi * fs. where fs is the sampling frequency. We generally only concerned the spectrum of 0-pi, because, according to the law of Nernst Naike, only the signals within the range f = fs / 2 is the effective signal to be sampled. Then, w in the range, the resulting spectrum is certainly about n / 2 is symmetric.

For example, if made 16 points FFT analysis, the maximum frequency f your original analog signal = 32kHz, the sampling frequency is 64kHz, the range is 0 ... 15 n. (Lu Note: This means that the original analog signal has been sampled 8 times.) At this time, the analog frequency 64kHz is divided into 16 points, each one is 4kHz, this is called frequency resolution (Lu Note: do with FFT the more points, the higher the frequency resolution). Then the abscissa, n = 1 f is the time corresponding to 4kHz, n = 2 corresponds to 8kHz, n = 15 is the time corresponding to 60kHz, the frequency spectrum you are symmetrical about n = 8. You only need to care about n = 0-7 within the spectrum is enough, because, had the highest frequency of the analog signal is 32kHz.

Here are two possible conclusions.

· First, the original sampling frequency fs must know how much of the signal, can know the exact frequency of each n is the number corresponding to the calculated actual frequency of the k-th point is f (k) = k * (fs / n)

· Second, after you make a 64kHz 16-point FFT, since the frequency resolution is 4kHz, if the original signal at 5kHz or 63kHz weighty, you can not be seen in the spectrum, which means they want you painted spectrum realistic, it is necessary to take more points to do the FFT, n is greater, you have to take on a longer time-domain signal samples to do the analysis. But in any case, because the principle of discrete sampling, you can not be completely and accurately draw the original continuous real time spectrum of the signal, only infinitely close to (n is infinite time), this is called frequency leakage. Under no circumstances become the sampling frequency fs, the frequency of leakage can be improved by taking more points, it can also be improved by adding a window before doing FFT, that's another topic.

Portal

References paper format

Wherein for books, paper documents focus on the precipitation, which identifies the type of document recommends one letter "A"; for other unspecified Document Types recommended single letter "Z".

We can learn about specific references format

[No.] journal authors. Title [J]. Title. Publication year, volume (period): beginning and ending page.

[No.] monographs author. Title [M]. Edition (the first edition can be omitted). Publication: Publisher, year of publication: beginning and ending page.

[No.] Proceedings of. Title [C]. editor. Proceedings name. Publication: Publisher, year of publication: beginning and ending page.

[No.] dissertation authors. Title [D]. Save Location: preservation units, year.

[No.] patent owner. Patent Document Title [P]. Country: patent number. Release date.

[No.] standard number, standard name [S] of Publication: publisher, year.

[No.] of the newspaper. Title [N]. Newspaper name, date of publication (edition).

[No.] report's authors. Title [R]. Reports to: report organizer year.

[No.] of electronic documents. [Electronic Document Title carrier type identification]. Journal of date.

Portal

---

Portal 2

Feedforward

Adjusting the feedforward input using open loop control system or directly acting control signal (feed forward signal) constituting the disturbance.

When the control section sends a signal, the instruction is not issued controlled portion of the feedback signal, but rather feed-forward signal before sending after receipt of a stimulus from the monitoring means, applied to the control section, so as to make early adaptive response mediated controlled promptly moving parts.

The reaction lag feedforward control system can avoid the negative feedback regulation overkill fluctuation generated in the reaction and, to control the adjustment faster and faster. Illustration, person before a race, although the signal gun has not sounded, before the body through the feed regulation, the contestants have appeared heart rate, cardiac output, increased pulmonary ventilation.

Portal

How word 2019 add subscripts

Superscript: Select to set the upper corner of the target number, hold down the Ctrl Shift and "=" sign;

Subscripts: To select the subject of digital set lower corner, hold down Ctrl and "=" sign.

Portal

How word 2019 table settings dotted line

Select - Borders and Shading - Select and access lines



Portal

Fourier transform matlab program settings

Portal

Fold matlab code

Home - Preset - Code Folding - sections:

Portal

Wavelet transform function

wavedec wavelet decomposition
waverec wavelet reconstruction
appcoef approximate coefficients
detcoef detail coefficients

Portal

#wavedec function meaning of L and C
C is a column vector, which is stored in the wavelet detail coefficients obtained after decomposition of the respective layers and approximate coefficients CDi last level CA. 5 layer is decomposed, for example, their storage structure, C = [CA5; CD5; CD4; CD3; CD2; CD1].

L is a column vector, which is stored in the length of each set of coefficients C, the layer is decomposed to Example 5, L = [len5; len5; len2;; len4; len4 len1]. Knowing this, the coefficients can be combined thresholding as C, then call a signal reconstruction waverec

wrcoef ( 'type', C, L, 'wname', N)
in type is a or d, or a specified portion of the image mean details. c is an array, which is stored in the coefficient of each component of the wavelet transform, (an, dn, dn- 1, ..., d2, d1). L is an array, the length of each recording element in C, (lengthof (an), lengthof (dn), ..., lengthof (d2), lengthof (d1)). wname is the name of the wavelet function is used to determine the reconstruction filter, N is the beginning of reconstruction of the first layers, that is, to point out to reconstruct for which coefficient.

Portals to the chiefs bloggers

waverec and the difference #wrcoef
wavedec is wavelet decomposition, the decomposition of a signal into the specified number n, and returns the wavelet coefficients of each layer.

waverec-- wavedec contrast with its role, i.e. a given wavelet coefficients to reconstruct the signal completely disposable.

This input is wrcoef-- wavelet coefficients, reconstructed signal. But with some differences above, except that it is the original signal reconstructed at the specified level, high-frequency or low-frequency components. In other words, this signal is not the original signal, but on approaching a certain level.

Portal

# Denoising threshold function
ddencmp P222 obtain default soft and hard threshold
thselect P224 denoising threshold selection
wbmpen P225 return penalized threshold
wdcbm P227 obtaining threshold
wpbmpen P229 return penalized threshold
WDEN P231 automatic denoising
wdencmp P234 denoising
wpdencmp P239 denoising
wpthcoef P241 thresholding
wthcoef p242 thresholding
wthcoef2 P243 thresholding
wthresh p244 hard and soft thresholding

# Wavelet family

About wavelet reconstruction

Excellent bloggers

matlab Forum

Baidu answer

matlab wavelet transform de-noising

Reprinted outstanding bloggers: https: //blog.csdn.net/Arrogant_95/article/details/80744388

All Clear; CLC
Load ( 'Audio_1_resample.mat');
S = data_resample;% acquiring a signal to be processed, data_resample .mat above is a parameter in the
% of the total signal length of
N = numel (S);
% wavelet decomposition;
[c, l] = wavedec ( s, 7, 'coif5');% wavelet basis as coif5, decomposition layer 7 layers
ca11 = appcoef (c, l, 'coif5', 7);% acquiring low-frequency signal
cd1 = detcoef (C, L,. 1);
of Cd2 = detcoef (C, L, 2); obtaining the high frequency detail%
CD3 = detcoef (C, L,. 3);
CD4 = detcoef (C, L,. 4);
CD5 = detcoef (C, L,. 5);
a CD6 = detcoef (C, L,. 6);
CD7 = detcoef (C, L,. 7);
SD1 = zeros (. 1, length (CDl));
SD2 are = zeros (. 1, length ( cd2));% 1-3 0,4-7 layer facing layer is treated with a soft threshold function
SD3 = zeros (. 1, length (CD3));
SD4 = wthresh (CD4, 'S', 0.014);
SD5 wthresh = ( CD5, 'S', 0.014);
SD6 = wthresh (a CD6, 'S', 0.014);
SD7 = wthresh (CD7, 'S', 0.014);
= C2 [CA11, SD7, SD6, SD5, SD4, SD3, SD2 are, SD1];
S0 = waverec (C2, L, 'coif5');% wavelet reconstruction
Figure;
the subplot (211); Plot (S); the subplot (212); plot (s0) ;% Paint

Specifically decomposition level, the selection of which layer denoising function, which layer or for other operations is set to 0, which is selected or determined according to Wavelet signal to be processed, to analyze specific issues.

Portal

Looking minimum points matlab

Signal is negated - Find the maximum value and the maxima with findpeaks

[pks,locs]=findpeaks(-1*dp,'minpeakdistance',100)  

%%dp是数字信号；'minpeakdistance'设置两峰值间的最小间隔数

The complete code is as follows:

[pks,locs]=findpeaks(-1*dp,'minpeakdistance',100)
% subplot(212),
plot(locs,pks,'*');title('极小值点');grid on;

---

%% 基本用法
[pks,locs]=findpeaks(-1*dp)

Portal-to-excellent blogger

write formulas word 2019 shortcut keys

Hold down the alt, then =; that is,alt+=

Equation Editor can be generated as follows:

sort array a column matlab-based, and other columns moving along

Portal

Based on the second row in descending order:

a2=sortrows(a,2,'descend')

Example shows

a=[2,114;3,666;1,222]
a1=sortrows(a,2)  %默认升序
a2=sortrows(a,2,'descend')  %降序

Cluster analysis

Portal

coordinate range setting matlab plot of

Use codeaxis([0,2000,-300,700])

Example:

figure,
plot(d(:,1),d(:,2),'*'),axis([0,2000,-300,700]),title('所有幅值');

Sample Entropy

Code Portal

Matrix Code Portal

Portal
Portal 2

Hilbert transform and spectral analysis

Portal comes hht MATLAB function on
the portal using MATLAB syntax hht

Portal

sparse matrix

Hilbert transform concepts

Portal

Subscript word when writing the formula

Portal

word to bring up the ellipsis

1, win10 own input method is switched to Chinese
2, shortcut keys shift+6(input or Chinese slh)