Hi Guys 

Seeing as AI systems are all the rage right now and going mainstream and becoming easier to setup, I have been thinking about going down the machine learning rabbit hole with my inverter 😉  and seeing if I can use one for the control of my inverter battery settings so as to optimise the usage of my batteries and minimise my use of grid power and negate me having to be changing my settings all the time!

Using Load Shedding schedules, forecast solar radiation & historical time of day power usage, my thinking is that it should be able to reasonably accurately forecast the amount of battery power that the inverter needs to keep in reserve for the forecast conditions and adjust the System Mode settings to best optimise the setting for minimum grid power usage. 

I have also been thinking of using a system for the management of large loads on the inverter based on load shedding status & forecasts, time of day forecasted solar radiation, current solar production, battery soc. 

Anyone have any experience with Machine Learning platforms? I was thinking of trying out the Google Cloud platform, primarily because it is free to try out so will only cost me some time to see if I can figure out how it works. 

https://cloud.google.com/automl 

Is there anyone else working/thinking about a system like this? 

Edited March 13, 20233 yr by Sc00bs

Found this on the Home Assistabt Forum  https://community.home-assistant.io/t/taking-home-assistant-to-the-next-level-home-assistant-assistant-ai-powered-machine-learning-for-ha/294593/31 

 

Looks pretty interesting

I really believe this is the next step with domestic power management. Weather based power generation like wind and sun on a utility scale already has very stable and accurate predictor models. I'll follow with interest.  

On 2023/03/13 at 9:40 AM, Sc00bs said:

Hi Guys 

Seeing as AI systems are all the rage right now and going mainstream and becoming easier to setup, I have been thinking about going down the machine learning rabbit hole with my inverter 😉  and seeing if I can use one for the control of my inverter battery settings so as to optimise the usage of my batteries and minimise my use of grid power and negate me having to be changing my settings all the time!

Using Load Shedding schedules, forecast solar radiation & historical time of day power usage, my thinking is that it should be able to reasonably accurately forecast the amount of battery power that the inverter needs to keep in reserve for the forecast conditions and adjust the System Mode settings to best optimise the setting for minimum grid power usage. 

I have also been thinking of using a system for the management of large loads on the inverter based on load shedding status & forecasts, time of day forecasted solar radiation, current solar production, battery soc. 

Anyone have any experience with Machine Learning platforms? I was thinking of trying out the Google Cloud platform, primarily because it is free to try out so will only cost me some time to see if I can figure out how it works. 

https://cloud.google.com/automl 

Is there anyone else working/thinking about a system like this? 

I was just thinking about this now and Googled to see if someone was or is thinking of an ai automated system that configures time of use settings on the inverter automatically as load shedding schedules change..

 

Is there any progress on this tech?

15 hours ago, oddMan said:

I was just thinking about this now and Googled to see if someone was or is thinking of an ai automated system that configures time of use settings on the inverter automatically as load shedding schedules change..

 

Is there any progress on this tech?

Nothing on the AI front as yet that I have seen @oddMan 

I did post a video on adjusting your  battery settings based on your system solar power generation forecast for the next 24hrs. 

Works very nicely but would be nice to have some load shedding resilience built in as well.