Utility Meter outliers

Configuration
1

Hi,
I set up a Utility Meter for one of my plugs and it works fine most of the time. However, sometimes, Utility Meter shows outliers. I think they might be related to float rounding or something.

image
image1920×1077 103 KB

image
image1920×1077 100 KB

This is really not nice.
My first question is how to fix this. If I go to Deverloper Tools → Statistics and select this Meter, I can only find a value for every 5 minutes, but since this outlier is at an odd time, I cannot find and fix it.

Second question: How to avoid this in the future?

Thanks in advance!

2

Show the source sensor history for the same period.

3

Sure, here it is

image
image1920×1077 97.2 KB

Also, for reference, here’s the power chart.

image
image1920×1077 107 KB

4

Can you zoom in around 5:05 pm on the energy sensor?

5

It should still be in one of the 5 minute periods.

6

Just click the Outliers button - works for me.

7

Here are the zoomed in versions of the Utility Meter and the source sensor.

image
image1920×1077 90.5 KB

image
image1920×1077 96.7 KB

image
image456×684 25.5 KB

Not for me, it will show me some values with ~0.2 kWh. Not really outliers.

image
image453×832 31.8 KB

8

Ok so the glitch is not in the statistics data, only the state data. Good but very odd. Are you sure you had not corrected the statistics already?

Good news is that the state data will be deleted after your purge_keep_days interval.

Bad news is that I have no idea why the state data is showing that.

9

Yes, I’m sure haven’t changed anything.

As said, I think it’s a rounding issue. The value before is 7.069 kWh, then the outlier at 7,068.977706158489 kWh and then 7.069 kWh again

10

The utility meter sees that decrease as a reset. So when the next value comes in the whole 7.069kWh is added to the total.

Why is your sensor decreasing?

11

I don’t think, it’s decreasing. You see the sensor values in the screenshot above.

Also, if the utility meter would see that as a reset, I think we should see a step instead of a spike. so the values on the right should also be increased

12

Check the values of the energy sensor very carefully around the time the glitch occurred. Even a 0.001 decrease is bad.

13

How would I see such small decreases? In the plot above, the value before is 215 Wh, the value at the glitch timestamp is 215 Wh and the value after is 215 Wh. No decimals shown

15

I think, I found a hint. At that time, I must have done an update.

Entries from the logbook
image
image

Unlikely a coincident