finding slope of a signal

Hello. I have two signals (according to figure) and I want to find the slop of each signal.

Thanks for further help.

[jonas]

Well, as you can see the slope is varying. What exactly are you looking for? The trend? The change in every point?

[Sara]

sorry for the poor explanation. I want to get the trend.

Answer 1

[Dimitris Kalogiros]

A script that calculates mean slope :

clc; clear; close all;

%%generate input signal
x=0:1:1500;
y=(-0.002*x+3)+abs(sin(2*pi*.002*x))+0.2*sin(2*pi*.01*x);

figure; plot(y, '-b'); zoom on; grid on;
xlabel('x'); title('input signal y'); hold on;

%%first order aproximation
p=polyfit(x,y,1);

polyValues=p(1)*x+p(2);

plot(polyValues,'--r');
legend('input signal', 'first order approximation');

%%slope detection
fprintf(' mean slope = %f \n', p(1));

Of course, I had to generate a signal similar to what you presented at your question.

If you run the script you will receive the following:

[jonas]

Note that you can replace

polyValues=p(1)*x+p(2);

by

y = polyval(p,x)

So you don't have to write out the equation.

[Sara]

Dear Dimitris,

Thanks for your help. The red line is what I need. But is there any way to get more accurate slop line? I mean is it possible that the end point of red the line will be the same as the end point of the signal.

[jonas]

You can draw a line from the start point to the end point, but what's the point of that? It's certainly not indicative of future values.

[Dimitris Kalogiros]

@Sara , If the end point of the red line and the end point of the signal are the same , then you might have a significant error on what you have calculated as "mean slope". I believe, the method I used for the calculation of the "red line" is (probably) the best one.