heady forge
There is a known bug on APSystems EZ1 Micro-inverters for solar panels, where the 2 separate internal values for energy production will reset to 0 after reaching 540.6 kWh
(The bug has been officially confirmed by APSystems devs and should be fixed in newer 1.9 firmware)
https://github.com/SonnenladenGmbH/APsystems-EZ1-API/issues/34
The issue here is that it causes a big spike in the energy graphs but not in the actual statistics so they cant just be fixed in the database
In this screenshots, you can see Panel 1 was reset on 24th January and Panel 2 a few days after that
When Panel 1 goes from 540.6 to 0 KwH, the Total (P1+P2) goes from
13h00 = 1066 total (540.6 + 525.5)
14h00 = 525.7 total (0.05 + 525.65) -> This is the big 525.7 yellow spike in Energy graph
When Panel 2 reset:
13h50 = 555.93 total (15.32 + 540.61)
13h51 = 15.32 total (15.32 + 0) -> This is the smaller 15.32 yellow spike in Energy graph
here's the SolarTotal Lifetime Production on 24th January in the database
lusty sentinel
I don't follow your claim. The energy graphs are derived directly from the statistics.
Any spikes in the graph should be fixable via statistics adjustment.
idle crane
I agree.
Also, a resetting sensor (to 0) should not cause an issue if it has state_class: total_increasing. If it's set to state_class: total I would actually expect a spike downwards.
lusty sentinel
His problem is trying to sum resetting sensors. When one resets, the sum ends up with a big decrease, but not to zero. That's not suported by any state class.
idle crane
Ah, yeah, that's not possible indeed. Then you need to adjust the statistics manually indeed.
An option could be to add a utility meter for every PV sensor. And make that handle the reset. Then you can safely sum the utility meters as they would not return to 0 again.
Or let the energy dashboard do the summing. Thus add them all individually to the dashboard
heady forge
His problem is trying to _sum_ resetting sensors. When one resets, the sum ends ...
Actually, the micro-inverter is already doing the sum, it's supplying 3 "lifetime production" sensors, 2 for the P1 and P2 and a third for the combined P1+P2
These 2 sensors are total_increasing (src: https://github.com/home-assistant/core/blob/dev/homeassistant/components/apsystems/sensor.py#L63)
lusty sentinel
Is there still a question?
heady forge
Yeah sry, the thing that confuses me is I know how to fix glitchy spikes but in this case I dont understand what the "abnormal values" UI is going to do and am afraid it will mess things further.
It does identifies the 2 spikes, 525.88 and 15.34 but looking at the database, there is no such values...
Only the "stats" field which is just a copy of the inverter sensor values
And a "sum" field which seems to be where the spikes comes from as the difference in 2 successive values gives exactly the 525.88 spike (1568.88 - 1043)
So if use the UI and tell it 0 kWh instead of 525.88 kWh, is it gonna directly write 0 in that "sum" field ? that doesn't sound correct, because then the graph will display 0 - 1043 = 1043 kWh ?
Ideally it should change all the values after the first wrong, so that all the substractions are correct for all subsequents values... is this what the code is going to do ?
lusty sentinel
Ideally it should change all the values after the first wrong, so that all the substractions are correct for all subsequents values... is this what the code is going to do ?
Yep it does exactly that
change the 525.xx to 0 and you'll be good.
(and the 15 too also to 0, if that's incorrect as well)
The only thing energy cares about is the sum column. The adjustment process makes the updates to the sum column.
heady forge
It worked perfectly, thanks a lot ! @lusty sentinel
sacred igloo
Hello, I think I have the same issue: I have a big (unrealistic) spike on the data on January 31st, that makes the stats unreliable, can you explain how I can fix it as if I were 5 years old? I'm still learning technical maintenance on Home Assistance 😅
lusty sentinel
Hello, I think I have the same issue: I have a big (unrealistic) spike on the da...
developer tools / statistics, for each statistic w/ this issue, click the adjust sum button on the right of the row, click 'outliers', locate the high peaks, change them to 0.
sacred igloo
developer tools / statistics, for each statistic w/ this issue, click the adjust...
Thank you!
I'm wondering if there is a way to make an automation that would do this for each outliers found?