how to filter a noisy data from sensor
Hi, i have noisy data graph from sensor, but i still can see the trend looking like a sine wave. i implemented a extended kalman filter but it gives me a wrong trend . Does EKF can smooth the signal?
Asked by dmngu9 on 2015-12-09 23:17:37 UTC
Comments
If you expect a sine wave you could do a Fast Fourier Transform and cut the higher frequencies that would correspond to the noise. This works if your sine wave doesn't have a very high frequency. So just try a basic low pass filter.
Asked by Mehdi. on 2015-12-10 02:53:48 UTC
The thing is i want it to smooth the data in real time . if i do what you suggest, would it guarantee real time performance
Asked by dmngu9 on 2015-12-10 19:00:23 UTC
It is called "FAST" Fourier transform for a reason... Digital signal processing is really fast when it comes to filtering and you should definitely try out some stuff.
Asked by Mehdi. on 2015-12-11 07:23:27 UTC
i have problem with deriving state transition equations for ekf. does ekf requires precise transition model equation
Asked by dmngu9 on 2015-12-13 05:03:18 UTC