peak average of consecutive values in a matrix array

2023 7 19
1 个回答
回答已采纳
更新时间：2023 7 19
31 次查看（30 天）

0 个投票

Hello, I would like to calculate the average of the peaks (positive and negative values) of consecutive values from a predifined number of elements from a Nx1 matrix, where N is the number of rows. For example, let's say I have a matrix with the following form (24x1):
A[1;2;3;4;5;6;7;8;9;10;11;12;13;14;15;16;17;18;19;20;21;22;23;24]
I want to obtain the peak and the average for every 5 elements,
B=[mean(peak(1,2,3,4,5)); mean(peak((2,3,4,5,6)); mean(peak((3,4,5,6,7));...........mean(peak((20,21,22,23,24))]
I am chosing every 5 elements to simplify the example. But, this number number of elements can be higher than 100 and the points vary dramatically from positive to negative as they come from acceleration earthquake signals from N higher than 245000.
I would appreciate the help.

6 个评论
显示 4更早的评论 隐藏 4更早的评论

Dyuman Joshi
Dyuman Joshi 2023-7-19
Following up from your previous questions, are movmean and movmax not working for you?
Hi, yes it worked. But peaks estimations involve as many peak values (positive or negative) not just one when using movmax. Thank you.
Dyuman Joshi
Dyuman Joshi 2023-7-19
"But peaks estimations involve as many peak values (positive or negative) not just one when using movmax."
Can you provide an example for this?
Of course. I have attached an example file where there are three columns. Each column represents the component X, Y and Z of the acceleration earthquake signal in cm/sec2. I could say based on the previous description I would like to estimate the average of the peaks (positive and negative) for every 51 elements from the total signal that contains 9600 points for each component of the signal X, Y and Z. Thank you for your support.
Dyuman Joshi
Dyuman Joshi 2023-7-19
Ran in:
Ran in:
Suppose this is the data in hand -
x = randi([-20 20],1,15)
x = 1×15
-5 -16 14 -18 19 -3 18 1 -10 -6 -12 16 5 8 0
Assume the moving window is 4, what should be the output in thise case? and what is the logic behind it?

请先登录，再进行评论。

请先登录，再回答此问题。

 采纳的回答

Mathieu NOE
Mathieu NOE 2023-7-19

0 个投票

hello
maybe this ?
I am using peakseek , this fex submission is available here
PeakSeek - File Exchange - MATLAB Central (mathworks.com)
, but you can use the standard (slower) findpeaks function offered by TMW
the code provided below will buffer your data with length = 51 samples , then split the signal for each cahnnel into a positive and negative signal and find the positive and negative peaks , then take the mean of them (separately not mixing pos and neg peaks !)
example for X channel :
data = readmatrix('Signal.txt'); % acceleration earthquake signal in cm/sec2 ,X, Y and Z
[samples, channels] = size(data);
dt = 1; % sample rate (s)
t = dt*(0:samples-1); % time vector
%% home made solution (you choose the amount of overlap)
buffer_size = 51; % how many samples
overlap = 0; % overlap expressed in samples
%%%% main loop %%%%
[new_time,data_out_pos,data_out_neg] = my_peakmean(t,data,buffer_size,overlap);
figure(1),
plot(t,data(:,1),new_time,data_out_pos(:,1),'*-g',new_time,data_out_neg(:,1),'*-r');
title('X');
legend('raw data','pos peaks mean','neg peaks mean');
xlabel('Time(s)');
ylabel('cm/sec2');
figure(2),
plot(t,data(:,2),new_time,data_out_pos(:,2),'*-g',new_time,data_out_neg(:,2),'*-r');
title('Y');
legend('raw data','pos peaks mean','neg peaks mean');
xlabel('Time(s)');
ylabel('cm/sec2');
figure(3),
plot(t,data(:,3),new_time,data_out_pos(:,3),'*-g',new_time,data_out_neg(:,3),'*-r');
title('Z');
legend('raw data','pos peaks mean','neg peaks mean');
xlabel('Time(s)');
ylabel('cm/sec2');
%%%%%%%%%% my functions %%%%%%%%%%%%%%
function [new_time,data_out_pos,data_out_neg] = my_peakmean(t,data_in,buffer_size,overlap)
% NB : buffer size and overlap are integer numbers (samples)
% data (in , out) are 1D arrays (vectors)
shift = buffer_size-overlap; % nb of samples between 2 contiguous buffers
[samples,channels] = size(data_in);
nb_of_loops = fix((samples-buffer_size)/shift +1);
for k=1:nb_of_loops
start_index = 1+(k-1)*shift;
stop_index = min(start_index+ buffer_size-1,samples);
x_index(k) = round((start_index+stop_index)/2);
% find peaks
for n = 1:channels
tmp = data_in(start_index:stop_index,n);
tmp_pos = tmp;
tmp_pos(tmp_pos<0) = 0;
tmp_neg = tmp;
tmp_neg(tmp_neg>0) = 0;
[locsp, pksp]= peakseek(tmp_pos,1,max(tmp_pos)/10); % pos peaks above 10% of max pos amplitude
[locsn, pksn]= peakseek(-tmp_neg,1,max(-tmp_neg)/10); % neg peaks above 10% of max neg amplitude
data_out_pos(k,n) = mean(pksp); % mean of pos peaks
data_out_neg(k,n) = mean(-pksn); % mean of neg peaks
end
end
new_time = t(x_index); % time values are computed at the center of the buffer
end

5 个评论
显示 3更早的评论 隐藏 3更早的评论

This is a very clever answer straight to the issue. I really appreciate your help.
I was checking the code, it works perfectly fine. However, the estimation of peak mean (defined as 51 as a buffer for this example) should go after each consecutive points. In other words, the averaging should start in time 26, then 27, 28, 29 ...until 9575. This means that the averaging of peaks (positive and negative) are estimated from the following range of points:
for time 26 the averaging of peaks are estimated from points 1 to 51
for time 27 the averaging of peaks are estimated from points 2 to 52
for time 28 the averaging of peaks are estimated from points 3 to 53
for time 29 the averaging of peaks are estimated from points 4 to 54
. .
. .
for time 9575 the averaging of peaks are estimated from points 9550 to 9600.
This can be handle with the overlap value set to 0?
I would appreciate your help
Best
Mathieu NOE
Mathieu NOE 2023-7-19
hello again
thanks for the nice comments
to get the required behaviour simply change the overlap value , you need now :
overlap = buffer_size-1; % overlap expressed in samples
Mathieu NOE
Mathieu NOE 2023-7-19
just one point : in my code , the time vector is zero based
so the first buffer (points 1 to 51) , assuming a sampling rate = 1 , correspond to t = 0 to 50 , so my buffer is centered at t = 25 (and not 26 as in your comment above)
this explain why we have a time difference of 1 between you and me.
my new time vectors is 25,26,27,... 9574
I really appreaciate your help. it is sorted now. Thank you very much.

请先登录，再进行评论。

更多回答（0 个）

类别

在 帮助中心 和 File Exchange 中查找有关 Descriptive Statistics 的更多信息

产品

版本

R2021a

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by