How to find the slope of phase data (between two points) which is uneven

I have a phase vs frequency plot. I need to find the slope between point 'a' and point 'b' in an automated way instead of looking at the points 'a' and 'b' and calculating the slope. Can anyone help me with this?

Answer 1

[Image Analyst]

You ask ONLY about the slope, not about finding a and b, so I assume you've found "a" and "b" already. So, assuming your "a" is index1 and "b" is index2, you can do this

coefficients = polyfit(x(index1):x(index2), signal(index1:index2), 1);
slope = coefficients(1);

[Sriram Guddati]

That's what, I have to find the points a and b in an automated way, instead of looking at the graph and figuring them out.

[Image Analyst]

OK, that wasn't stated in the original question. So you can use

[peakValues, indexOfPeaks] = findpeaks(signal);
[valleyValues, indexOfValleys] = findpeaks(-signal);
valleyValues = -valleyValues; % Flip again to get right side up.

Of course you will find lots of peaks and lots of valleys so you'll have to decide which to concentrate on.

[Sriram Guddati]

Thank you, I will try this.

Answer 2

[jonas]

I would suggest using either findpeaks to determine points a and b and then calculate the average slope or alternatively using findchangepts. If you want code, then I suggest you upload some data.

[Sriram Guddati]

Thanks for the suggestion. I will try to find a and b using findpeaks. And is there a way to find the indexes of a,b using findpeaks function

[jonas]

Yes of course. The index is the second output of findpeaks. Note that you need to run the function two times, one for peaks and one for valleys. For the latter, you just turn the plot upside down, i.e.

findpeaks(-y)

[Sriram Guddati]

Thanks, I will try this method

Answer 3

[Greg Dionne]

You could try a combination of unwrap (to remove the jumps of your array) and sgolay to do the differentiation. See the second example on how to perform differentiation.