That could happen, but not likely. I am using open source api’s that exist to share weather knowledge in the world.
That is probably caused by the live day-time training. When it was cloudy today at your site this has issues. I’m ironing those issues out as we speak. Still finetuning solar forecast to be perfect…
And yet again a new test version 8.16.0: https://homey.app/a/com.gruijter.powerhour/test/
I was not happy with how the day-time learning on cloudy days was messing up the solar learning profile. So I again spent a HUMONGOUS amount of time rethinking and rewriting the solar forecasting algorithm. I tested the algorithm by creating virtual solar panel system all over the world and in different times of the year. And I am very content with the result (if I say so myself 
).
I have good hopes this will be my last rewrite of solar forecasting, although the proof of the pudding will have to come from different panels, in different parts of the world, in different times of the year!
So, in case some of you find the forecaster often being way off, you can now send your data to me for analysis. I created a maintenance button for that. After using that button you must manually create an app diagnostics report, and put your panel orientation in the message (e.g. south. east or west). I will then receive your data and can start investigating. Note that on cloudy days the forecast will always be off on an hourly basis, but the average over the entire day should still be ok.
My own forecast for today (intermittent / partly clouded):
The distribution graph shows how your panels are modeled by the PbtH forecast learning. The dotted line shows how well the model fits real measured data. Above 60% can be considered good. So in my graph it shows that the forecaster works (very) well between around 9:30 - 20:30. Having multiple sunny days in the past 2 weeks significantly improves the modelling.
Looks like the device didnt fully migrate. Try restarting the app or rebooting Homey.
Bothe not successful.
Pls send an app diagnostics report
f0b560e3-c1ec-40e3-94fe-3ccc40124fde
App version 8.16.1 was just released as stable: Power by the Hour | Homey
@Martin_Koning Please verify if this fixes your graphs issue.
Yes, both issues solved.
90% accuracy at the edges
12.00 - 16.00 modelfit 65%
also have a hanger here… now rebooting homey as rebooting the app does not help.
define hanger. You mean the images? Are you using v8.16.1? And/or did a reboot solve it?
prices were not send to the EMS, just some random hours, but after reboot of the homey it was solved. I also changed the EMS to app2app comms which seems to work great!
Is there a reason why Model Fit before 08.00 en after 20.00 is 0%?
A good model 06.00 - 08.00 and 20.00 - 22.00 is also logic ico predicting battery-needs.
It is just a statistical thing. Doesnt mean the prediction is useless at dusk and dawn. When a prediction says 200W at 8am, but in reality it was 100W, the model is 100% ‘off’, while stil accurate within 100W.
Maybe an idea,
In the “battery” option, it calculates its benefit with charged energy per tarif and uncharged energy per tarif. (Dynamic rates)
But if you use solar only. The tarif of charging could be zero…
Maybe ad an option to select if you want to use the charging or uncharging or both
Well, that depends on your view. I setup PbtH to keep track of the ‘benefit’ of your solar panels in a solar summarizer. So if you ALSO set the charging tariff in the battery summarizer to 0, you are actually counting the same solar benefit twice.
So the battery summarizer uses grid tarifs as ‘cost’ during charging. And at the same time the solar panels uses the same grid tariff as ‘benefit’. IMO that is the most clean way of doing the bookkeeping.
I understand. But now the battery goes negative so it “costs” charging by tariff. While it actually charged on solar…
Yes, and you can add it with the earnings from solar. So total always is good.
But if you really want to, just make an extra dap that always has 0 as price, and change the active dap group the battery uses on the fly via a flow.
One thing is confusing for me. Our energy company moved from “hour” to “15 minute rates” so I’m changing all the flows. For example: where I first had “The price is 1 of the 5 lowest in the next 8 hours” I expected I could change that to “The price is 1 of the 20 lowest in the next 8 hours” because I moved to “15 minute rates” and 20 x 15 minutes would equal 5 hours. But the limit that I can enter is still 5. Does it still look at hours? Or is it something that was overlooked?
