A new release has just been deployed, v0.5.949. The main headlines for this release are: - Added solar forecast yield display for today/tomorrow to dashboard - Keep existing agile slots in place and recombine with auto-population logic, which will result in user selected agile slots not being overwritten at 16:35 More details to full release notes here: https://community.wonderwatt.com/d/96-wonder-watt-release-notes

Just seen the solar forecast on the dashboard. I guess at the moment, this is just for information only rather than adjusting Agile charging. Are there any correction factors that can be set to make the forecast more accurate baased on my actuals?

> @"johalareewi"#p6211 this is just for information only rather than adjusting Agile charging For info only, yes. We have a correction factor in the background but not exposing that yet, as it would likely lead to users tweaking it too often and generally ending up with more outliers than otherwise.

Hi, glad to see the work on solar forecasting. Wondering if it you have historical data and could publish e.g the total estimate month by month so we could compare with data from our inverters. I'd be interested to see how constant a correction factor would be across a whole year.

@"Ianmcarthur"#p6224 We will in due course. We have only started collecting this systematically very recently. We're aiming to start collecting actual PV yields too, so we can create personalised corrections on that basis. Or even use a combination of last rolling 14 days and weight it, then combine with weighted daily forecast.

That sounds even better. Many thanks for the advance info.

Just for clarity, it is only the pre-selected evening slots that are kept. I tried selecting slots for the next day before the re-populate time and they were overwritten.

@"WindyMiller"#p6236 Hmm, shouldn't be, we did test that scenario too. We'll try and reproduce. Nope, just ran it for our own account after clearing preselected slots, and saving two that were not the lowest price. After running prepopulation, the two previously selected ones remain. What agile charge window have you got set up?

@"admin"#p6239 elaborate-routine 7 charge slots between 22:30 and 06:00 with hold set for the rest. Then I unset one and set 3 more leaving me with the 9 you can now see. I did this around 4.20pm and checked back at 5pm to find they were reset so set them again. I'll check again tomorrow.

@"WindyMiller"#p6242 One thing to note is that we effectively keep the agile slots you have selected and saved, and the auto-population changes get applied on top of that. So if you unselect for instance one of the cheaper slots that would be autoselected, it will be autoselected again.

@"admin"#p6245 That might explain the one I unselected but not the other three. Let me try again tomorrow to see what happens. Perhaps a flag against each slot to show whether it has been touched by the user or not?

Release 0.5.949 - Solar forecast plus Agile slots overwrite fix

admin

WindyMiller Nope, nothing particular in relation to this, beyond the previous explanation. ;-)

kerregan

admin I’ve been monitoring actual solar generation against the forecast for several days.

Here’s the results: forecast / actual

8.7/2.1, 4.6/2.1, 3.4/0.3, 10.6/4.7, 3.8/0.3, 4.5/1.7, 6.1/1.4, 3.8/0.1, 3.8/0.1

As you can see there is little correlation. I don’t believe there is anything wrong with my panels or inverter. The panels are clean, both strings are generating. My location, orientation and roof pitch are set up correctly.

I know it is very early days but I am wondering if you have seen similar discrepancies from other users. Perhaps accuracy will look a lot better when we get some actual sunshine!

terrible-guidance

admin

kerregan Thanks for that, yeah, at these lower numbers, the accuracy is lacking. Once we collect actuals and create personal correction skews, this should hopefully look a lot better.

« Previous Page