1
\$\begingroup\$

I want to build a system for smart grid which can optimize the grid power, load and battery charging to reduce my costs for grid power. For the optimization problem, I must know the current date to optimize the overall system, so the problem for me is how can I predict the solar power to optimize the system for the next hour. I want to know if there is a controller algorithm that will predict the power from a solar panel and then decide the power to system for optimization.

In my matlab simulation, I implemented a clear-sky model that estimates the solar insolation according to the altitude angle (Chapter 7 of Gilbert Masters' book Renewable and Efficient Electric Power Systems ) . By using this insolation data, I want to predict the power in real hardware and then allow the controller to decide how to operate for the next hour, depending upon the previous hour's values. This insolation data that I have calculated is not predicted (it is estimated); we don't know the weather condition for the next day, or even the next hour.

What algorithm or controller must I have to use to predict the solar insolation for use in a real-time hardware simulation depending upon the weather conditions?

\$\endgroup\$
12
  • \$\begingroup\$ So there is no secondary source of power? Just the panel itself? In my area it can go from solid overcast to clear skies in less than an hour. (We have a giant wind tunnel called the Columbia Gorge nearby.) You'd need a lot of information to make predictions here. \$\endgroup\$
    – jonk
    Commented Mar 1, 2017 at 18:45
  • \$\begingroup\$ i have grid another source of electricty that i mentioned above. \$\endgroup\$
    – Fame313
    Commented Mar 1, 2017 at 18:53
  • \$\begingroup\$ I was reading your desire to predict and focused on that. But yes, I see the answer to my first question there. Thanks. \$\endgroup\$
    – jonk
    Commented Mar 1, 2017 at 18:56
  • \$\begingroup\$ Dear @jonk ,Actually i want your decision . if you know please help me \$\endgroup\$
    – Fame313
    Commented Mar 1, 2017 at 19:09
  • 3
    \$\begingroup\$ "controller must decied how to operate in next hour depending upon the previous hour value" - why? \$\endgroup\$
    – Bruce Abbott
    Commented Mar 1, 2017 at 19:36

1 Answer 1

Reset to default
1
\$\begingroup\$

Short term prediction of solar irradiation can be tricky, especially in cloudy days.

For daily average based on a year there are software like PVSyst which is probably the most advanced solar plant forecasting software out there, used by major banks for plant forecast.

Running a simulation with this software will give you a very good starting point but won't account for daily climate, which I guess you could use weather forecast and refine the daily power simulation given by the software, with empiric method or why not some machine learning.

\$\endgroup\$
3
  • \$\begingroup\$ Dear @Damien ,, Thanks for answering to my question . this is good software . i try to use it . can you tell me that will it not give me online predictio of solar power. i mean i have to optimize the power for next hour depending upon the decision of privious hours. \$\endgroup\$
    – Fame313
    Commented Mar 2, 2017 at 12:28
  • \$\begingroup\$ Dear @Fame313, As said before short term prediction can be tricky, if the day is plain sunny then it's relatively easy as you can pretty much know how much irradiation you should receive for any given position, if it gets cloudy, rainy, or any type of overcast then you will need to work with local sensor. This is a big question recently raised for Smart Grid and gird balancing and has no simple answer, weather forecast, satellite imagery and irradiation sensor around the plant with good software is probably the closest it can get in our days \$\endgroup\$
    – Damien
    Commented Mar 8, 2017 at 3:26
  • \$\begingroup\$ You can get real time temperature (maybe pressure humidity) data from other smartphone users in your region. This can be used to predict data. Some apps and startups are already trying to do this. But this may not be accurate enough to predict small deviations. \$\endgroup\$
    – Indraneel
    Commented Oct 8, 2018 at 16:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.