MATLAB: Extract data at regular intervals

extracting data

Hi There,

I have 2 columns of data, time and concentration. The time data are at intervals of about 5 seconds and there is a corresponding concentration measurement at each time. There are about 20,000 time measurements and I want to reduce this to a more manageable amount. How can I obtain say every 10th time x concentration value?

Thanks

Best Answer

Are you asking for just

X = ones(2e4,2);
Y = X(1:10:end,:);

Y is a two-column matrix containing every 10-th measurement (or measurements taken every 50 seconds).

Related Solutions

MATLAB: Pointer Setting for identifying time of corresponding data

Let's see if I have your situation right.

Your measurements are stored in a 9450x10 matrix where each row is a row corresponds to a measurement time and each column corresponds to a device.

Your time stamp is a 105 x 1 cell array where each element is a string containing the timestamp associated with 9 measurements made for each of the 10 devices during that hour. So each time stamp is associated with 90 different measurements.

You want a way of looking at something like measurement 12 on device 8 and figuring out that the time stamp is '0001H' .

In that case, you can associate each row of measurements with a time stamp by creating a 945x1 time stamp array. If your initial time stamp array is called baseTsArray, and you make 9 measurement per hour:

measurementsPerHour = 9;
expandedTsArray = reshape( repmat( baseTsArray ,  1 , measurementsPerHour)' , [] , 1);

The new time stamp array would look like this:

    expandedTsArray =
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0001H'
    '0002H'
    ...
    '0105H'
    '0105H'
    '0105H'
    '0105H'

The time stamp associated with data(n,D) would be given by expandedTsArray{n} .

MATLAB: Does the PDE toolbox yield negative values for the concentration, when I solve a diffusion equation

The diffusion equation is a parabolic equation. The stability and accuracy of the solution is affected by the size of the time steps. The inaccuracies in the temporal integration lead to unphysical results for the concentration. This can be fixed by reducing the time steps to a smaller value.

I have scaled the original equation such that the temporal integration occurs in units of days. The scaling factor has been absorbed in the diffusion constant C (or Deff). Now we can divide the time range into 20,000 steps except that the size of the step is smaller than the original case, where it was 1 unit.

The solution for the issue is to change the diffusion constant so that it is integrating in units of days instead of seconds.

This can be accomplished by doing the following steps in the PDE model:

1. Rescale the solver time range from:

linspace(0,72000,72000) to
linspace(0,1,20000)

2. Modify the "C" parameter of the equation to account for the scaling

8E-17/(24*3600)

This resulted in non-negative values for the concentration.

Other reasons that could explain negative values for the concentration include numerical gradients that might arise due to poor mesh quality in certain regions of the domain.

Best Answer

Related Solutions

MATLAB: Pointer Setting for identifying time of corresponding data

MATLAB: Does the PDE toolbox yield negative values for the concentration, when I solve a diffusion equation

Related Question