Repository navigation
Agile Predict Integration #3240
|
I've been struggling to get this to work, but given that Agile Predict data exists, how can I get it to use that data instead of just copying the previous days slots? Tried a few things like using: custom_buy_price_sensor: sensor.predbat_agile_predict_bridge and pointing metric_octopus_import to something but either it just ignores it or doesn't compute a plan. Anyone else managed to get agile predict (or similar?) prices into predbat? |
Replies: 3 comments 10 replies
|
This was asked for earlier in the week #3231 At the moment there is no easy way to get Agile predict rates into predbat, there simply isn't the API or entities to provide that data that predbat reads. You are aware that you can use the Norddata configuration into Predbat https://springfall2008.github.io/batpred/energy-rates/#future-agile-energy-rates which gives pretty good forecast of what Agile prices will be for the next day. The prices arrive around 10am and are available for both Agile Import and Export (unlike Agile Predict which is import only). I have found this is more than good enough for Predbat when I have been on Agile, it predicts the future Agile prices for 11pm and overnight. Not as far in the future as Agile Predict but most of us are running our batteries on a daily cycle so knowing what will happen that night is good enough |
Yes, predbat duplicates today's rates into tomorrow if they are not available yet. Then if the norddata market data is configured in apps.yaml, at about 10am, Predbat will populate tomorrow's Agile prices (import or export) with the electricity market prices which are a pretty close approximation of what the actual Agile prices are going to be - ahead of the normal 4pm+ price release. |
|
Definitely feels like there's some value in predictions still instead of relying on duplicated data. To add some context on use case, I have a PHEV that doesn't necessarily need charging fully every day. If I delay charging an extra day it will cut charge prices by 75%, which is exactly the type outcome that gives predbat value As it stands currently though, predbat thinks both days have the same cost from the rate duplication |
Only for tomorrow's prices, and only up to 10am, then the norddata prices appear in the plan. For most people the default predbat plan is 24 hours ahead so this isn't a major limitation. I agree though, having the predicted prices in predbat would give a longer term view for those on Agile import. There's no equivalent price prediction for Agile export though as far as I know. |
|
Managed to get a POC working merging real rates with agile predict api data, capped at end of next day. I have only briefly tested the plan shows the data, so this may have bugs / edge cases Set in predbat Apps.yaml: metric_octopus_import: 'sensor.predbat_octopus_agile_rates'Sensor to pull from agile predict api: - platform: rest
resource: https://prices.fly.dev/api/D
scan_interval: 3600
name: Agile Predict
unique_id: agile_predict
value_template: "{{ value_json[0]['name']}}"
json_attributes_path: "$[0]"
json_attributes:
- "created_at"
- "prices"Template sensor to use confirmed next day data, or fallback to prediction - sensor:
- name: "agile_next_day_or_predict"
unique_id: agile_next_day_or_predict
state: "ok"
attributes:
rates: >
{% set real = state_attr(
'event.octopus_energy_electricity_<YOUR_ACCOUNT_HERE>_next_day_rates',
'rates'
) %}
{% set current = state_attr(
'event.octopus_energy_electricity_<YOUR_ACCOUNT_HERE>_current_day_rates',
'rates'
) %}
{# Determine last end time of current day rates, if present #}
{% if current %}
{% set last_current_end =
(current | map(attribute='end') | map('as_datetime') | max)
%}
{% else %}
{% set last_current_end = none %}
{% endif %}
{# CASE 1: true next-day rates exist #}
{% if real %}
{% set ns = namespace(out=[]) %}
{% for r in real %}
{% set ns.out = ns.out + [{
"start": (r['start'] | as_datetime).isoformat(),
"end": (r['end'] | as_datetime).isoformat(),
"value_inc_vat": r['value_inc_vat'],
"is_capped": r['is_capped']
}] %}
{% endfor %}
{{ ns.out }}
{# CASE 2: convert agile_predict to same structure, excluding overlap #}
{% else %}
{% set pred = state_attr('sensor.agile_predict', 'prices') %}
{% if not pred %}
{{ [] }}
{% else %}
{% set midnight =
now().replace(hour=0, minute=0, second=0, microsecond=0)
+ timedelta(days=2)
%}
{% set ns = namespace(out=[]) %}
{% for p in pred %}
{% set dt = p['date_time'] | as_datetime %}
{% set end_dt = dt + timedelta(minutes=30) %}
{# Exclude anything overlapping current_day_rates #}
{% if dt < midnight and (last_current_end is none or dt >= last_current_end) %}
{% set ns.out = ns.out + [{
"start": dt.isoformat(),
"end": end_dt.isoformat(),
"value_inc_vat": ((p['agile_pred'] | float) / 100) | round(5),
"is_capped": false
}] %}
{% endif %}
{% endfor %}
{{ ns.out }}
{% endif %}
{% endif %}Template sensor that takes either confirmed rates from octopus integration, or merges agile predict in for missing slots from above template sensor, reformatted into predbat's expected data format - sensor:
- name: "Predbat Octopus Agile Rates"
unique_id: predbat_octopus_agile_rates
unit_of_measurement: "£/kWh"
state: >
{% set now_ts = now() %}
{% set current = state_attr(
'event.octopus_energy_electricity_<YOUR_ACCOUNT_HERE>_current_day_rates',
'rates'
) %}
{% set ns = namespace(val=none) %}
{% if current %}
{% for r in current %}
{% set st = r['start'] | as_datetime %}
{% set en = r['end'] | as_datetime %}
{% if now_ts >= st and now_ts < en %}
{% set ns.val = r['value_inc_vat'] %}
{% endif %}
{% endfor %}
{% endif %}
{% if ns.val is none %}
{% set pred = state_attr('sensor.agile_predict', 'prices') %}
{% if pred %}
{% for p in pred %}
{% set dt = p['date_time'] | as_datetime %}
{% set end_dt = dt + timedelta(minutes=30) %}
{% if now_ts >= dt and now_ts < end_dt %}
{% set ns.val = ((p['agile_pred'] | float) / 100) | round(5) %}
{% endif %}
{% endfor %}
{% endif %}
{% endif %}
{{ ns.val if ns.val is not none else 0 }}
attributes:
raw_today: >
{% set current = state_attr(
'event.octopus_energy_electricity_<YOUR_ACCOUNT_HERE>_current_day_rates',
'rates'
) %}
{% if current %}
{% set ns = namespace(out=[]) %}
{% for r in current %}
{% set ns.out = ns.out + [{
"start": (r['start'] | as_datetime).isoformat(),
"end": (r['end'] | as_datetime).isoformat(),
"value": r['value_inc_vat']
}] %}
{% endfor %}
{{ ns.out }}
{% else %}
{% set pred = state_attr('sensor.agile_predict', 'prices') %}
{% if not pred %}
{{ [] }}
{% else %}
{% set today_end =
now().replace(hour=0, minute=0, second=0, microsecond=0)
+ timedelta(days=1)
%}
{% set ns = namespace(out=[]) %}
{% for p in pred %}
{% set dt = p['date_time'] | as_datetime %}
{% set end_dt = dt + timedelta(minutes=30) %}
{% if dt < today_end %}
{% set ns.out = ns.out + [{
"start": dt.isoformat(),
"end": end_dt.isoformat(),
"value": ((p['agile_pred'] | float) / 100) | round(5)
}] %}
{% endif %}
{% endfor %}
{{ ns.out }}
{% endif %}
{% endif %}
raw_tomorrow: >
{% set rates = state_attr('sensor.agile_next_day_or_predict', 'rates') %}
{% if not rates %}
{{ [] }}
{% else %}
{% set ns = namespace(out=[]) %}
{% for r in rates %}
{% set ns.out = ns.out + [{
"start": r['start'],
"end": r['end'],
"value": r['value_inc_vat']
}] %}
{% endfor %}
{{ ns.out }}
{% endif %}
created_at: "{{ now().isoformat() }}" |
|
very impressive, thanks @Mazo Keep testing it, and once you have a verified version it'd be great to include this in the predbat documentation for others to use |
|
It is possible to adapt an agile tariff using an automation into the expected Octopus format. I'm doing it for Amber into Predbat |
|
@hnb2907 how did you get on? Is it working? (I'm thinking of doing the same) |
This was asked for earlier in the week #3231
At the moment there is no easy way to get Agile predict rates into predbat, there simply isn't the API or entities to provide that data that predbat reads.
You are aware that you can use the Norddata configuration into Predbat https://springfall2008.github.io/batpred/energy-rates/#future-agile-energy-rates which gives pretty good forecast of what Agile prices will be for the next day. The prices arrive around 10am and are available for both Agile Import and Export (unlike Agile Predict which is import only).
I have found this is more than good enough for Predbat when I have been on Agile, it predicts the future Agile prices for 11pm and overnig…