Automation to detect high power and run automation but ignore spikes

[mark.carline (Mark)]

I want to run an automation when my EV car charger starts and then stops. The trouble is that there are times when it can spike for 1 minute which i want to ignore. The sensor gets updated every 30 secs.

I have built these two automations, the first one works great for when power increases:

- id: '1725743193929'
  alias: EV Socket is using more than 2kW (for 1 min)
  description: Detect when EV charging has started
  trigger:
  - platform: numeric_state
    entity_id:
    - sensor.sdm120_active_power
    above: 2000
    for:
      hours: 0
      minutes: 1
      seconds: 0
  condition: []
  action:
  - action: notify.mmyemail_gmail_com
    metadata: {}
    data:
      message: EV Socket is using more then 2kW at {{now().strftime("%H:%M %Y-%m-%d")}}
        active power is {{ states('sensor.sdm120_active_power') }}
      title: EV Socket is using more then 2kW at {{now().strftime("%H:%M %Y-%m-%d")}}
        active power is {{ states('sensor.sdm120_active_power') }}
      target: [email protected]
  mode: single

but this one runs for every spike:

- id: '1725787889752'
  alias: EV Socket is using less than 1W (for 1 min)
  description: Detect when EV charging has finished
  trigger:
  - platform: numeric_state
    entity_id:
    - sensor.sdm120_active_power
    for:
      hours: 0
      minutes: 1
      seconds: 0
    below: 1
  condition: []
  action:
  - action: notify.myemail_gmail_com
    metadata: {}
    data:
      message: EV Socket is using less than 1W at {{now().strftime("%H:%M %Y-%m-%d")}}
        active power is {{ states('sensor.sdm120_active_power') }}
      title: EV Socket is using less than 1W at {{now().strftime("%H:%M %Y-%m-%d")}}
        active power is {{ states('sensor.sdm120_active_power') }}
      target: [email protected]
  mode: single

Is there a way to have a single automation which has an action when it is detected and then runs when the condition returns to normal ?

[zigomatic]

@mark.carline

sensor.sdm120_active_power

There is no spike when it’s below 1 W. It’s no power but I guess there is still some power. Check how your value is moving in the history and adapt the min. power.

Like your car never takes less then xxx W when charging.
I would put below 100W

[Rofo (Ro)]

If the actual question is, ‘can i combined this into a single automation?’, then from what I can see, this looks pretty straightforward.

Add both triggers to a single automation and give each a trigger id.

Then use an if or choose condition to run the appropriate action, based on the trigger id.

[aceindy]

Have a look at Filter - Home Assistant

[mark.carline (Mark)]

thanks @aceindy @Rofo @zigomatic

I think what i meant to say was:

• There is a scenario where the car gets plugged in and it draws power for 1-2 minutes and I do not want any automation to run at all (on the up “leading edge” or on the down “trailing edge”).

• The key senario i want to be able to run two automations on is:
  a) When it starts (and it has been drawing above 2000 watts for 1-2 minutes)
  b) When it finishes (when it returns to less than 1W).

Thanks in advance

ok ?

[zigomatic]

@mark.carline

There is a scenario where the car gets plugged in and it draws power for 1-2 minutes and I do not want any automation to run at all (on the up “leading edge” or on the down “trailing edge”).

The question is how to identify that scenario is happening ?
It’s clear for me that you need to check history of the value and identify which part you want to receive a notification and which one you don’t want.

And as proposed, not use the real value but the filtered value where you can remove what you don’t want to see.

[Rofo (Ro)]

So, I think what you are saying is that in the first 1-2 minutes there might be a spike of energy above 2000watts and you want to ignore that, you just want it to trigger when the power draw is at a steady stead above 200watts ?

If this is the case, then this should be fairly straightforward. You could extend this:-

    for:
      hours: 0
      minutes: 1
      seconds: 0

To be (3 minutes), but you might miss a minute or two of the steady state (not sure how consistent the power draw pattern is).

Alternatively, use something like the (low pass) filter as linked above, some other examples of how to use a filter are here:-

Filter values average every 5 minutes

No, sensor.pm25_average is what I’d expect your existing code to generate: If the filter platform creates different entity IDs than the normal way of just slugify-ing the name, then use that entity ID instead. Yes, that, as I said:

[mark.carline (Mark)]

Thanks to everyone for trying to help. I thought that this would do the trick:

sensor:
  - platform: filter
    name: "SDM120 Active Power Time Simple Moving Average (2 min)"
    entity_id: sensor.sdm120_active_power
    filters:
      - filter: time_simple_moving_average
        window_size: "00:02"
        precision: 2

BUT !

The source entity (sensor.sdm120_active_power) is from Modbus and that only seems to create a new datapoint in HA whenever there is a change. So any kind of filter based on average data isn’t going to work.

Any more ideas?

EDIT 1: Looks like somebody else had this issue and no reply:

Can I add a trigger to a sensor "filter" platform to force it to update periodically?

This works reasonably well - platform: filter unique_id: "-------------------" name: "Foo" entity_id: sensor.bar filters: - filter: time_simple_moving_average window_size: "00:10" precision: 1 It’s mostly ok but it does have a tendency to get “stuck” at a high value when the underlying sensor drops back down to 0 and never changes. As with a lot in HA, this update optimization works fine in most cases but for sensors that are a function of time (moving avgs, derivative…

EDIT 2: I also tried “force_update: true” as per this thread and doenst work in modbus:

Lowpass filter does not recalc if sensor value remains unchanged

Lowpass filter only does adjust/recalc if the sensor value changes. If the sensor returns a stable value (lets say solid 0% or 100%) then the value remains at whatever it was at the last change. - platform: filter name: "boiler average" entity_id: sensor.solar_boiler filters: - filter: lowpass time_constant: 3 precision: 0 How can I trick the lowpass filter to recalc even on stable input so its value approaches solid 0% (or 100%) in the example. see the blue averaged (low…

EDIT 3: Also tried to do a force refresh via this for both the modbus entity and the filter entity and no impact:

Home Assistant Core Integration

[mark.carline (Mark)]

Just an update - i completely failed to get anything working in automation or filters due to the lack of datapoints in a time window. It would seem that the sdm120/modbus integration returns zero then there no more datapoints in HA that can be used (it can be like that for hours if the value doesn’t change).

I did manage to get a template binary_sensor time working great, this is my final solution:

template:
  - binary_sensor:
      - name: "EV Charging Socket"
        unique_id: "EV Charging Socket"
        device_class: battery_charging
        delay_on:
          seconds: 60
        delay_off:
          seconds: 60
        state: "{{ states('sensor.sdm120_active_power') | float > 2 }}"

Thanks everyone for helping.