How to remove Spikes from a signal

Hello everyone,

I have a plot below which consists of two curves:

1. Red Curve - Original

2. Green Dotted Curve - Target

I want to filter out the spikes from the 'Original Curve' and replace it with suitable values such that the 'Original Curve' appears like the 'Target Curve'. In the plot below, I have achieved it using Interpolation. I want to find an easier way to do it. I even tried using 'medfilt1' function but I am not able to get rid of all the spikes. Any help would be highly appreciated.

Thanks !!

[Ankit]

I will first find out the reasons why I am getting these spikes and answers to following question

is this behavior expected or not?

Whether it is because of Solver settings? Is my time time step is too large?

If you real want to avoid this behavior have a look on below links:

https://de.mathworks.com/help/signal/examples/signal-smoothing.html

https://de.mathworks.com/help/curvefit/smoothing-data.html

[Mani Teja]

Hello Ankit,

Thanks for your reply. I would like to mention that I am aware of the reasons due to which the plot has spikes. My only aim is to get rid of the spikes in the simplest way possible and make it look like the 'Target Curve'.

[Ankit]

hi Mani

great can you please attached your complete plot instead of part of it?

thanks

ankit

[Mani Teja]

Hello Ankit,

the attached image is the complete plot. Please scroll towards right to see the complete plot.

Thanks

Ankit

[Ankit]

sorry my bad!

have you tried sgolayfilt() also?

here you can find more examples for filtering out signals

https://de.mathworks.com/matlabcentral/answers/354402-how-to-remove-spikes-from-the-signal

https://de.mathworks.com/matlabcentral/answers/356713-remove-spikes-from-a-signal

[Ankit]

I forgot to ask you which MATLAB version you are using?

[Adam]

I'd be surprised if a smoothing filter could help you convert that to your target. Spikes generally considered as being anomalous to the main trend, but in your case it's the majority of your data, at least from -100 to -60 that is 'spikes' and almost none of the data in that range is down at your target line and none of it below the target.

[Mani Teja]

Hello Ankit,

thanks for the links and the function. I shall try using them. I am using Matlab 2017b currently.

[Mani Teja]

Hi Adam,

Please scroll towards right to see the complete plot.

Thanks

Mani

[Ankit]

Hi Mani,

I was just going through some options in MATLAB, go to know about the below options

Clean Outlier Data: https://de.mathworks.com/help/matlab/ref/cleanoutlierdata.html

I found one such example using filloutliers (see below link)

filloutliers| isoutlier | rmoutliers

Regards

Ankit

[Mani Teja]

Hello ankit,

The 'filloutliers' function helped precisely.

thanks :D