Integration via trapezoidal rule in various sections of the same array

FW さんによって質問されました 2019 年 10 月 22 日

最新アクティビティ Jon さんによって編集されました 2019 年 10 月 22 日

16 ビュー (過去 30 日間)

If the signal S consists of several peaks as a function of time, is there a way to assign integration limits for the trapezoidal rule? Say, I would like to integrate peak 1 from 5 to 7 min, then there is another peak 2 which needs to be integrated from 7.5 to 9.5 minutes. The documentation for trapezoidal rule does not mention anything about limits. Assume we do not know the functional form of the peaks. Thanks.

https://www.mathworks.com/help/matlab/ref/trapz.html

2 件のコメント

darova on 22 Oct 2019

You can manually divide/split your data in specific areas. Use findpeaks to find peaks

FW on 22 Oct 2019

Yes, sure but the key question is what should be the syntax? For example findpeaks gives the peak position [peaks, index] and peaks have a bseline width. If my signal is S, how should integrate if the peaks are very close, say the the time index for peak 1 is 150:200, and the other peak is adjacent with some overlap so its index is 200:350. How would one apply trapz for selected ranges? Thanks.

サインイン to comment.

1 件の回答

リンク

Answer by Jon on 22 Oct 2019

Edited by Jon on 22 Oct 2019

Accepted Answer

Assuming you can find the indices of your ranges by some means, e.g. findpeaks, then make an array of the ranges, for example

ranges = [103 128;
          182 196;
          297 315]

then loop through the ranges using the trapezoidal rule for example

numRanges = size(ranges,1)
F = zeros(numRanges,1); %preallocate array to hold integrals for each range
% loop through ranges
for k = 1:size(ranges,1)
    F(k) = trapz(Y(ranges(k,:))) % assuming your original signal is in vector, Y
end

2 件のコメント

FW on 22 Oct 2019

Thanks, I will try this approach. These indices in the range will belong to the Y values (or Signal values) not the indices corresponding to time values, is this correct? I use the following syntax to find peaks [ peaks index] =findpeaks(Y); With the help of index values I find the values of corresponding time ranges.

Jon on 22 Oct 2019

Suppose you have a vector of time values, let's call it t, and a vector of corresponding y values, let's call it Y. Suppose using findpeaks or some other function you determine a range of interest in your vector y starts at index values which we will call r1 and r2 respectively. Then you can find the corresponding y values using Y(r1:r2) and the corresponding time values using t(r1:r2).

Note if you use the function findpeaks then you should call it using the syntax

% get location and width of peaks
[pks,locs,w] = findpeaks(Y);
% calculate half width of peaks as an integer count of indices 
halfWidth = round(w/2);
% calculate start and end indices for each peak
r1 = locs - halfWidth % vector of starting indices
r2 = locs + halfWidth % vector of ending indices
% put the ranges into a matrix for further use
ranges = [r1(:) r2(:)] % colon makes sure that vector is a column

サインイン to comment.

サインインしてこの質問に回答します。

現在この質問をフォロー中です

FW (view profile)