DESS consumption forecast to high because EV charging

Hello,

i’ve noticed that the consumption forecast in VRM is much to high when i have charged my EV in previous days.

I do automatically enable my wallbox when grid current is cheap. I do not charge my EV from the Battery. I dont have an intelligent wallbox, it is simply enabled from a shelly when price is below threshold.

First screenshot showing when i charged my EV on Wednesday 1st of January:

Second screenshot showing the forecast for tomorrow Wednesday 22nd of January:

Clearly to see is that the charging consumption in the history influences the forecast. Which is to be expected when the forecast relies on historic data, but this causes real trouble in my situation because the price is not cheap in this timeframe tomorrow and the DESS tries to cover this from battery…
Also the average consumption of other days is much too high. It seems as if this big charging event spreads out over the forecasts of even weeks afterwards.

Also occured for Friday 18th of January and 24th of Jan.


Possible solutions for me:

  • Is the a possibility to exclude certain consumption?
  • Could there be some correlation with the prices, eg. only take over high consumption in the forecast if prices where cheap an will be cheap again?
  • Exclude outliers (it almost seems like forecast for one hour is the mean over same hour at last N Mondays, maybe make it median?)

That’s right. I’ve made two observations in this context. First of all, the low price times repeat themselves relatively regularly, so I don’t think it’s a bad thing. The second effect that I’ve observed (2 wall boxes and 1 heat pump that charge directly at low prices), due to the slightly higher consumption forecast, I always have some reserve in the battery for unforeseen consumption. If there is really too much charging, the system usually pushes it to the most expensive times of the next day. This often becomes clear when the prices for the next day come. I have to mention in this context that I have a relatively large battery, so a little blocked capacity doesn’t really bother me. The whole situation, which I originally assumed would have a negative influence, is turning out to be an advantage more and more.

I have the same remark as Johann and also run my heatpump and ev charger on the lowest hours,
and it influences my system negatively.
VRM predicts the consumption during the night based on nights when the ev charger was running. It fills up the rest with battery charging to use during the peak times the next day. When the ev charger is not running that night, I could charge more from the grid, but VRM limits the charge to what it predicted before, even though it could charge more.

I have an alfen ev charger, but added it in the cerbo through mqtt. So it knows the ev consumption, I just have to be able to exclude that consumption from the VRM predictions. I could do the same for my heatpump which is also metered.

This would be also my favorite solution. Then i have complete data logging in VRM for later analysis but the forecast ist not influenced.
Workaround would be to send my grid meter readings (which come over MQTT) minus the power measured at the EV charger. But then VRM would not see the full energy consumption.

As per a previous post of mine I would still like to see the option for a User defined consumption profile.

That way a one off high consumption is not going to distort future forecasts. Also if I know that I may have a particularly high consumption the next day I can alter the profile accordingly.

At the moment DESS generally predicts a low consumption and I have to switch to ESS to avoid a flat battery or continually alter the desired minimum SOC to force a higher charge level and then adjust down during the day to meet demand.