Help interpolating irregular time series to regular time series

Hello,

I have irregular time series data: (time, X-acceleration, Y-acceleration, Z-acceleration)

19:24:26.850, -0.11492168, 0.038307227, 9.634268

19:24:26.859, -0.12449849, 0.038307227, 9.662998

19:24:26.872, -0.1340753, 0.05746084, 9.672575

19:24:26.884, -0.12449849, 0.02873042, 9.615114

19:24:26.889, -0.15322891, 0.047884032, 9.643845

19:24:26.902, -0.10534488, 0.06703765, 9.662998

19:24:26.909, -0.076614454, 0.076614454, 9.624691

19:24:26.921, -0.095768064, 0.05746084, 9.653421

19:24:26.935, -0.05746084, 0.095768064, 9.624691

19:24:26.939, -0.06703765, 0.05746084, 9.672575

etc.

How can I interpolate/resample(?) the data so that I can get accelerations every 0.010 seconds? I can add the date to the time string with no problems, I'll even have too soon. Thank you very much!

Oleg

[Jan]

Please explain the format of your data. Is "19:24:26.850" a string? Is this a cell array? When you post the data in a valid Matlab syntax, creating an meaningful answer is easier.

[Oleg]

I'm sorry, I'm very new to MatLab. Yes, it's a string from a log file that I read in like this:

fid = fopen('c:\MatLabData\log.txt', 'rt');

a = textscan(fid, '%s %f %f %f', ... 'Delimiter',',', 'CollectOutput',1);

fclose(fid);

format short g

M = [datenum(a{1}) a{2}]

RtSumSqsCol = sqrt(M(:,2).^2+M(:,3).^2+M(:,4).^2)

RtSumSqsColNoGrav = RtSumSqsCol-9.8

M = [M RtSumSqsColNoGrav]

to get data that looks like this:

   7.356e+05       -1.245       4.7884       11.281       2.5187
    7.356e+05      -1.1109       5.1427       11.435       2.7871
    7.356e+05     -0.99599       5.3439       11.358       2.7919
    7.356e+05     -0.74699       5.4779       11.262       2.7461
    7.356e+05     -0.74699       5.4779       11.262       2.7461
    7.356e+05     -0.73741       5.5641       11.253       2.7749

but I need it at a regular time interval. I will try the answers soon, thanks so far!

Answer 1

[Chad Greene]

Letting t be column 1 and x be column 2, interpolate values xi corresponding to ti:

ti = yourStartingTime + 0:.010:3; % this represents 3 seconds at 0.010 second intervals. 
xi = interp1(t,x,ti);

[Chad Greene]

If your time series is acceleration, spline interpolation may be more appropriate than the default linear.

xi = interp1(t,x,ti,'spline');

Answer 2

[Oleg]

Thank you for pointing me in the right direction! I was able to use your answer to get this:

Torig=M(:,1)

Acc = M(:,5)

plot(Torig,Acc)

datetick('x')

Ti=[Torig(1):1/24/60/60/1000:Torig(end)]';

Acci=interp1(Torig,Acc,Ti,'spline');

plot(Ti,Acci);

datetick('x')