# How to stitch together subsets of data from acceleration time histories?

2 views (last 30 days)
Kyle Tolbert on 27 Feb 2024
Commented: Kyle Tolbert on 28 Feb 2024
I have large time histories of data (in array) with the first colomn being time and the second column being acceleration (g). I need to find the peaks that are above a certain threshold and then create a subset of data that starts say 1 second before and ends 2 seconds after the peak. After that I need the subsets stitched together to create a "final" time history. Any help would be greatly appreciated!!!
Mathieu NOE on 28 Feb 2024
we could probably help if you would share the data (and some code if you have started something)

Voss on 28 Feb 2024
One thing you have to consider is what to do if a peak is within 2 seconds of another peak, so that the two peaks' 3-second-long intervals overlap: do you want to combine those intervals into a single interval (so that there's no data repeating in the final output), or do you want to extract the 3-second interval around each peak independently (so that there may be data repeated in the final output)? In the code below, I chose to combine overlapping intervals.
% parameters
peak_threshold = 0.6;
time_before_peak = 1;
time_after_peak = 2;
% random data
t = linspace(600,3600,30001);
g = 0.18*randn(size(t));
% g = 0.22*randn(size(t));
data = [t;g].';
% plot the data
plot(data(:,1),data(:,2))
xlim(data([1 end],1))
% find peaks above peak_threshold
% [p,idx] = findpeaks(data(:,2),'MinPeakHeight',peak_threshold); % if you have the Signal Processing Toolbox, you can use findpeaks
idx = islocalmax(data(:,2)) & data(:,2) > peak_threshold;
p = data(idx,2);
tp = data(idx,1);
% plot the peaks
hold on
plot(tp,p,'ro')
% construct the start and end times of each interval to extract
ts_te = tp(:)+[-time_before_peak time_after_peak];
% combine overlapping intervals:
idx = 1+find(ts_te(2:end,1) <= ts_te(1:end-1,2));
for ii = numel(idx):-1:1
ts_te(idx(ii)-1,2) = ts_te(idx(ii),2);
ts_te(idx(ii),:) = [];
end
% plot a green line to indicate the start of each interval
xline(ts_te(:,1),'g')
% and a red line to indicate the end of each interval
xline(ts_te(:,2),'r')
% extract the data between each start and end time into the cell array new_data
N = size(ts_te,1);
new_data = cell(N,1);
for ii = 1:N
start_idx = find(data(:,1) >= ts_te(ii,1), 1);
end_idx = find(data(:,1) <= ts_te(ii,2), 1, 'last');
new_data{ii} = data(start_idx:end_idx,:);
end
% combine all the intervals' data
new_data = vertcat(new_data{:});
% plot the extracted acceleration data vs extracted time
figure
plot(new_data(:,1),new_data(:,2))
xlim(data([1 end],1))
% or plot the extracted acceleration data vs its sample index (i.e.,
% eliminating the time gaps between extracted intervals)
figure
plot(new_data(:,2))
Kyle Tolbert on 28 Feb 2024
Thanks a ton, I will give it a try!

### Categories

Find more on Smoothing and Denoising in Help Center and File Exchange

### Community Treasure Hunt

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

Start Hunting!