Kalman filter

Hi all
I’ve been working on a new filter for the filter component, implementing a simple 1D Kalman filter which can be used to filter noise from a sensor readings. There is no perfect filter, and in practice you may need to try several different filters to get the desired result for your particular use case. So if you’ve not had success with the existing filters, you might want to try this one.

It would be really great to get feedback on how it performs in real world use, so if you are interested in trying it out and giving feedback please do, the repo is below.
Cheers

5 Likes

Hi

Is project alive?
I tried add it to hassio but it did not start
I add manifest.json

{
  "domain": "filter",
  "name": "kaman",
   "requirements": [],
  "dependencies": [],
  "codeowners": []
}

and empity init.py

added sensor

- platform: filter
  name: co2_kabinet
  entity_id: sensor.co2_kabinet_raw
  filters:
   - filter: kalman
     sensitivity: 0.8

But recive error on config check

Platform error sensor.filter - No module named 'custom_components.sensor.sensor'

It definitely needs an update, and welcome a PR :slight_smile:

hope to see updated version soon

btw i have good sensors for test in real life image

Have you tried the other filters? Moving average might be effective here

it have lagre time shift and regulation of vent is not perfect. Kalman filter reacts more quickly fo sudden transitions and is not sensitive to bounce

I just created a fork and did some changes, basically it works in my test env.

1 Like