how can I fit the sine curve for data having several periods?

hello. i am new to MATLAB. I have data set for which I have to create a sine curve. The original plot is quite fluctuating so creating a sine curve can lead to a average trend. I have attached the excel file in which original plot is already made. can somebody tell and show how to deal with this data? the period for the sine curve should be one year. thankyou

Answer 1

[Star Strider]

I have no idea what you want. Your data have one peak at a frequency of about 1, another with a frequency of about 37, and another one at the Nyquist frequency that is not usable.

The code the plot the fft:

[d,s] = xlsread('MUHAMMAD USMAN ARSHAD data.xlsx');

figure(1)
plot(d(:,1), d(:,2))
grid

Ts = mean(diff(d(:,2)));
Fs = 1/Ts;
Fn = Fs/2;
fd = fft(d(:,2));
fd(1) = 0;
fv = linspace(0, 1, length(fd)/2+1)*Fn;
iv = 1:length(fv);

figure(2)
plot(fv, abs(fd(iv)))
grid

[MUHAMMAD USMAN ARSHAD]

Thank you for your reply. actually this is hydrological data. this is change of groundwater table with the passage of time (years). In numerical modelling it is very difficult to simulate this behavior so my objective is to produce a sine curve for this data. this data is periodic in the sense because in every summers the peak goes up and in winters the peak goes down. I am just new to MATLAB. so i am not sure how it will work? I have attached an image that may help you in understand the objective of my question. the data is very scattered and i want to fit a sine curve..

[Star Strider]

That doesn’t look anything like the data in the file you posted, at least not to me.

Those data would be difficult to fit without filtering out the noise first, or isolating the two peak frequencies I found, one with a period of about a year and another with a period of about 10 days. (These are just guesses, since I don’t know what the time units are for the first column of data.)

However this technique for fitting sinusoidal data may suggest a way for you to do what you want.

[MUHAMMAD USMAN ARSHAD]

the units are in years for the time... this is just a image taken from the web for fitting sine curve for random data to show you that what i need..

[MUHAMMAD USMAN ARSHAD]

can you please guide me how i can do it? the data is very scattered so how i can simplify it? and add a sine fit for it

[Star Strider]

Nonlinear curve fitting is an iterative process, and the exact technique you end up using depends on your data. I would begin by subtracting the mean of your data from your data, then perhaps low-pass filter it to eliminate the noise, and then fit it with the technique I used in the link I provided in my previous Comment. I don’t know what parts of your data you want to fit and what is noise, so you have to decide how best to process it.

[MUHAMMAD USMAN ARSHAD]

ok thanks. i get back to you with these answers soon.. i will keep in touch thank you

[MUHAMMAD USMAN ARSHAD]

i will provide you the simplified data.. and post it after doing finding the answers to your question Sir..