Comments (11)
Hi thank you for the feedback.
Sounds like an interesting idea. And I did not find any similar library in a quick search..
I have also learned more about this topic since I wrote the post and I would suggest a slightly less complicated solution using scipy's filter state (zi
). One could do something like below, without the need to track past inputs and outputs, and solely relying on the scipy implementation:
zi = scipy.signal.lfilter_zi(b, a)
ys = []
for y in yraw:
yout, zi = scipy.signal.lfilter(b, a, [y], zi=zi)
ys.append(yout)
I intend to make an update to the post once I get around to it.
from yarppg.
Thanks @SamProell for this update.
If you are interested in Julia you might like JuliaDSP/DSP.jl#548
from yarppg.
Maybe some classes for stateful version of digital filters should be implemented
from yarppg.
sure, this was just a quick demonstration of how to use the filter state. Writing a class around this would be necessary.
from yarppg.
I tried
class StateFulLiveLFilter(LiveFilter):
def __init__(self, b, a):
"""Initialize live filter based on difference equation.
Args:
b (array-like): numerator coefficients obtained from scipy.
a (array-like): denominator coefficients obtained from scipy.
"""
self.b = b
self.a = a
self.zi = scipy.signal.lfilter_zi(self.b, self.a)
def _process(self, y):
"""Filter incoming data with standard difference equations.
"""
yout, self.zi = scipy.signal.lfilter(b, a, [y], zi=self.zi)
return yout
but MAE is a bit higher
lfilter error: 1.8117e-15 (with previous version)
lfilter error: 0.039018 (with this version)
from yarppg.
Same problem also occurs with
class StatefulLiveSosFilter(LiveFilter):
"""Live implementation of digital filter with second-order sections.
"""
def __init__(self, sos):
"""Initialize live second-order sections filter.
Args:
sos (array-like): second-order sections obtained from scipy
filter design (with output="sos").
"""
self.sos = sos
self.zi = scipy.signal.sosfilt_zi(sos)
def _process(self, y):
"""Filter incoming data with cascaded second-order sections.
"""
yout, self.zi = scipy.signal.sosfilt(self.sos, [y], zi=self.zi)
return yout
sosfilter error was 0 and is now 0.039018
from yarppg.
In principal, your implementation(s) are good. I played around with it myself and the "problem" lies with setting the initial state.
You should be able to see, that the results do not match only in the beginning of the signal.
Apparently, to reproduce the behavior of lfilter(b, a, y)
, zi
has to be initialized differently. The documentation for lfilter
states
If zi is None or is not given then initial rest is assumed. See lfiltic for more information.
After some playing around, I found that zi = scipy.signal.lfiltic(b, a, [0])
does the trick. At least for lfilter, I cannot find something similar for sosfilt.
from yarppg.
Ok so StateFulLiveLFilter
should probably be
class StateFulLiveLFilter(LiveFilter):
def __init__(self, b, a, yi=None, xi=None):
"""Initialize live filter based on difference equation.
Args:
b (array-like): numerator coefficients obtained from scipy.
a (array-like): denominator coefficients obtained from scipy.
"""
self.b = b
self.a = a
if yi is None:
yi = [0.0]
self.zi = scipy.signal.lfiltic(b, a, yi, xi)
def _process(self, y):
"""Filter incoming data with standard difference equations.
"""
yout, self.zi = scipy.signal.lfilter(b, a, [y], zi=self.zi)
return yout
no idea about how to overcome this sosfilt error.
from yarppg.
looks good. I started working on a "minor" rework of the internal structure, including formatting. If it's ok with you, I might just replace the current implementation of DigitalFilter
with this one. Or you can wait and then make a PR to get the full credit ;)
I'd say that having some deviation in the beginning is not that big of a deal for most applications. Focus is on online filtering, where some startup differences should be no problem. And also, you cannot get the same behavior with the original sosfilt anyway, i guess.
from yarppg.
Yes we can live with this initialization problem... but maybe that's a lack on Scipy side
I don't really care about the "full credit" 😄 Thanks @SamProell
from yarppg.
also many thanks to you!
from yarppg.
Related Issues (3)
- core dumped error in python3.11 HOT 3
- Video Input HOT 10
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from yarppg.