Random Spikes in Energy Dashboad

I have random spikes in the energy dashboard that i am unable to understand where they are coming from. They don’t have a pattern, usually occurring once-twice a day, sometimes not at all.

I am using a Solaredge Inverter with the ModBus Integration. It generates a lot of sensors. The ones that have the problem in the dashboard are:
sensor.solaredge_m1_imported_kwh
sensor.solaredge_m1_exported_kwh
They are counting the totals since the inverter was installed and never reset.

I did searched for similar threads here, in the github for core and for the integration but did not find anyone with this problem with the same

I looked at the sensor states but it does not have any unknown or 0 values, its always increasing.

When looking at the statistics and statistic_short_term in the sqlite file i do see the problem but there my ideas from where to look further stops. The states look good, it appears to me that the statistics code messes something up.

Any ideas where i can look further?

Check state_class

It says state_cass: total_increasing

I seem to be having the same issue, also with Solaredge integration (but I’m using Solaredge cloud…)

I’m seeing my SolarEdge cloud integration sometimes report spikes. Typically it is near the end of day, and the sensor reports a value equal to twice the energy produced up until that point in the day.

You can see this happened on April 20th and April 29th in the graph below. Graphs on monitoring.solaredge.com do not show the same spikes. Is the plugin possibly double counting production? Or is there a bug in how SolarEdge is reporting the values? What can I do to help debug this?

similar spikes in the “lifetime energy” value:

@randoms
check if your databases statistic table sometimes resets the state of the sensor to zero before going back to count up normally.
looks like this

state | sum
30     | 100
31     | 101
35     | 105
0      | 105
40     | 145
42     | 147

if thats the case you have a problem with the integration. it needs to report unavailable instead of 0 when solaredge does not provide data. the monitoring platform sometimes has short problems. you can switch to local monitoring via modbus to avoid this.

if you run HA on sqlite you can use the sqlite add-on to look into the database.