Open-Meteo Solar Forecast with multiple arrays

[daviessm]

I have compared solar forecasts for my location between Solcast and Open-Meteo and found that Open-Meteo tends to be more accurate. I'd like to use those forecasts in PredBat but I have it set up in HA with four Open-Meteo services, one for each string of panels.

Describe the solution you'd like
pv_forecast_[today|tomorrow|d3|d4] to take an array of sensors which are added together to give totals.

Additional context
If this already works, could it be documented instead? 😄

[gcoan]

The pv_forecast_today etc entites are designed to be used with the Solcast integration https://springfall2008.github.io/batpred/apps-yaml/#solcast-solar-forecast and I don't know that it works with open-meteo formatted entities, are the attributes the same as Solcast?
Have you tried?

[metabubble]

I also am using Open-Meteo, because of the sizes of my micro inverters, I can't set them up in solcast (they are under 1 MW).

As @gcoan said, Open-Meteo Solar Forecast is incompatible as it uses a different format for its attributes.

However, Gemini provided me with these template sensors to group the openmeteo ones together and turn them into a solcast-like format

  - sensor:
      - name: "Predbat PV Forecast Today"
        unique_id: predbat_pv_forecast_today
        unit_of_measurement: "kWh"
        # Sums the main state of all matching today sensors
        state: >
          {{ states.sensor 
             | selectattr('entity_id', 'match', '^sensor\\.energy_production_today(_[0-9]+)?$') 
             | map(attribute='state') 
             | map('float', 0) 
             | sum | round(2) }}
        attributes:
          detailedForecast: >
            {% set entities = states.sensor | selectattr('entity_id', 'match', '^sensor\\.energy_production_today(_[0-9]+)?$') | list %}
            {% set ns = namespace(totals={}) %}
            {# Aggregate all watts by timestamp #}
            {% for entity in entities %}
              {% set watts = state_attr(entity.entity_id, 'watts') %}
              {% if watts %}
                {% for time, w in watts.items() %}
                  {% set current = ns.totals.get(time, 0) %}
                  {% set ns.totals = dict(ns.totals, **{time: current + w}) %}
                {% endfor %}
              {% endif %}
            {% endfor %}
            {# Convert aggregated watts to 15-min kWh intervals #}
            {% set ns2 = namespace(forecast=[]) %}
            {% for time, w in ns.totals.items() | sort(attribute='0') %}
              {% set ns2.forecast = ns2.forecast + [{"period_start": time, "pv_estimate": (w / 4000.0) | round(4)}] %}
            {% endfor %}
            {{ ns2.forecast }}

      - name: "Predbat PV Forecast Tomorrow"
        unique_id: predbat_pv_forecast_tomorrow
        unit_of_measurement: "kWh"
        state: >
          {{ states.sensor 
             | selectattr('entity_id', 'match', '^sensor\\.energy_production_tomorrow(_[0-9]+)?$') 
             | map(attribute='state') 
             | map('float', 0) 
             | sum | round(2) }}
        attributes:
          detailedForecast: >
            {% set entities = states.sensor | selectattr('entity_id', 'match', '^sensor\\.energy_production_tomorrow(_[0-9]+)?$') | list %}
            {% set ns = namespace(totals={}) %}
            {% for entity in entities %}
              {% set watts = state_attr(entity.entity_id, 'watts') %}
              {% if watts %}
                {% for time, w in watts.items() %}
                  {% set current = ns.totals.get(time, 0) %}
                  {% set ns.totals = dict(ns.totals, **{time: current + w}) %}
                {% endfor %}
              {% endif %}
            {% endfor %}
            {% set ns2 = namespace(forecast=[]) %}
            {% for time, w in ns.totals.items() | sort(attribute='0') %}
              {% set ns2.forecast = ns2.forecast + [{"period_start": time, "pv_estimate": (w / 4000.0) | round(4)}] %}
            {% endfor %}
            {{ ns2.forecast }}

      - name: "Predbat PV Forecast D3"
        unique_id: predbat_pv_forecast_d3
        unit_of_measurement: "kWh"
        state: >
          {{ states.sensor 
             | selectattr('entity_id', 'match', '^sensor\\.energy_production_d2(_[0-9]+)?$') 
             | map(attribute='state') 
             | map('float', 0) 
             | sum | round(2) }}
        attributes:
          detailedForecast: >
            {% set entities = states.sensor | selectattr('entity_id', 'match', '^sensor\\.energy_production_d2(_[0-9]+)?$') | list %}
            {% set ns = namespace(totals={}) %}
            {% for entity in entities %}
              {% set watts = state_attr(entity.entity_id, 'watts') %}
              {% if watts %}
                {% for time, w in watts.items() %}
                  {% set current = ns.totals.get(time, 0) %}
                  {% set ns.totals = dict(ns.totals, **{time: current + w}) %}
                {% endfor %}
              {% endif %}
            {% endfor %}
            {% set ns2 = namespace(forecast=[]) %}
            {% for time, w in ns.totals.items() | sort(attribute='0') %}
              {% set ns2.forecast = ns2.forecast + [{"period_start": time, "pv_estimate": (w / 4000.0) | round(4)}] %}
            {% endfor %}
            {{ ns2.forecast }}

      - name: "Predbat PV Forecast D4"
        unique_id: predbat_pv_forecast_d4
        unit_of_measurement: "kWh"
        state: >
          {{ states.sensor 
             | selectattr('entity_id', 'match', '^sensor\\.energy_production_d3(_[0-9]+)?$') 
             | map(attribute='state') 
             | map('float', 0) 
             | sum | round(2) }}
        attributes:
          detailedForecast: >
            {% set entities = states.sensor | selectattr('entity_id', 'match', '^sensor\\.energy_production_d3(_[0-9]+)?$') | list %}
            {% set ns = namespace(totals={}) %}
            {% for entity in entities %}
              {% set watts = state_attr(entity.entity_id, 'watts') %}
              {% if watts %}
                {% for time, w in watts.items() %}
                  {% set current = ns.totals.get(time, 0) %}
                  {% set ns.totals = dict(ns.totals, **{time: current + w}) %}
                {% endfor %}
              {% endif %}
            {% endfor %}
            {% set ns2 = namespace(forecast=[]) %}
            {% for time, w in ns.totals.items() | sort(attribute='0') %}
              {% set ns2.forecast = ns2.forecast + [{"period_start": time, "pv_estimate": (w / 4000.0) | round(4)}] %}
            {% endfor %}
            {{ ns2.forecast }}

Just a heads up, there is currently an issue with 15 minute resolution of production which is addressed here.

And another heads up, it is currently unclear to me if it expects kW or kWh, as log messages with the current issue were misleading.

              {% set ns2.forecast = ns2.forecast + [{"period_start": time, "pv_estimate": (w / 4000.0) | round(4)}] %}

If once the bug is fixed it is off by a factor of 4, you might want to divide by 1000 instead of 4000.

[daviessm]

are the attributes the same as Solcast?

I thought they might be because HA can use Open-Meteo for solar predictions in the Energy tab, but obviously that works differently!

Have you tried?

No, because I'd also also seen this issue raised:

Just a heads up, there is currently an issue with 15 minute resolution of production which is addressed here.

So this can be closed unless there's any desire to add an Open-Meteo integration, because I don't really want to faff around with any more template sensors.

[gcoan]

So this can be closed unless there's any desire to add an Open-Meteo integration, because I don't really want to faff around with any more template sensors.

we can add it to the enhancement pile, but for now I suggest the template sensors are the way to go and I can add this to the documentation as an alternative to solcast

thanks @metabubble

And another heads up, it is currently unclear to me if it expects kW or kWh, as log messages with the current issue were misleading.

solcast pv today entities are energy values so should be kWh

[metabubble]

You're welcome. May I propose with waiting with the addition to the documentation until #3525 is fixed so we know if we divide by 1000 or 4000 to get fewer warnings in the log?