I used to be like OP. (have a similar background and have similar interests in tech for the planet)<p>Then I realised couple of things, an humbling experience:<p>1) given any position on earth, you can compute exactly what's the optimal inclination at any given point in time for a PV to maximize the energy production. Sure, there are reflection and secondary irradiation conditions (eg.: there is a lake close to it), but again, assuming the environment is static, it's way faster to just compute it statically rather than dynamically. Also, in most scenarios Beam irradiance from diffusion (the beam hitting the object) is order of magnitude higher than from reflective one (the same beam bouncing on a 3rd object first).<p>2) In mechanics movable part are the things to avoid. They have lower MTBF (mean time before failure) and as such they introduce complexity and increase cost<p>3) Economics is a key component of engineering. There is a cost to everything, the computational power, the energy needed by the servo, etc, etc. Given 1 and 2, a dynamic solution simply has a lower ROI than a static one.<p>I really appreciate the OP exploration here: there is a good overview of basic control theory and a good foundation of ML (although don't be deceived, this is a very simple modelling task that OP is overkilling with a way more complex model). That said, for everyone reading, this is not something you want to do in a real world situation.
I'll just leave this here...<p><a href="https://www.youtube.com/watch?v=wL9PcGu_xrA&t=174s" rel="nofollow">https://www.youtube.com/watch?v=wL9PcGu_xrA&t=174s</a>
I'd like to see a comparison with a dumb as a box of hammers controller like PI or extremum seeking on the local gradient. I can't imagine that it is learning much more than a simple strategy like this, but maybe that's my lack of imagination...
Doesn't this model fail to account for seasonal variations in the locus of the sun? The optimal angle will vary across the year, whatever the latitude.<p>Maybe I'm missing something, but i would use a simpler algorithm which doesn't need ML. On day 0, plug in the latitude and allow the system to traverse the range of angles, finding the optimal one at the time - ie: yielding maximum power. Let it run 3-5 times during the day, then fit those points to the theoretical path of the sun across the sky. Now your system is calibrated, without needing any other input. As the seasons change, the system will always know which angle to face for optimal power.
Sorry for the off-topic. I think if just 10% of the efforts were put on improving mini/micro/pico hydro electric generator rather than on solar systems, most of the rural areas will probably better off now with reliable power supply than relying solely on the intermittent solar power[1], [2].<p>[1]Micro hydro:<p><a href="https://en.m.wikipedia.org/wiki/Micro_hydro" rel="nofollow">https://en.m.wikipedia.org/wiki/Micro_hydro</a><p>[2]Micro hydro power with turgo generator:<p><a href="https://youtu.be/njNYuEKW-ek" rel="nofollow">https://youtu.be/njNYuEKW-ek</a>
I remember there was this science barge that had a solar panel that passively tracked the sun by being mounted on a metal tube that had some gas in it that expanded and somehow tilted the panel more towards the sun.<p>I wonder how much more accurate your system is and whether the tradeoff is worth the added expense of a motor + the additional maintenance cost of moving parts.<p>I wonder why solar farms don't use active tracking, is that added maintenance + equipment cost just not worth it?
As a small p.v. user (domestic, wired by myself): tracking Sun gives around 10-20% annual electricity production (witch is meaningless for self-consumption scenario) and a bit earlier and late electricity during the day (you produce earlier in the morning and later in the evening, witch is interesting for self-consumption) but beside the mere cost you need more space: fixed installs need just to count fixed shadows, not static of course, but easy to handle. Rotating means you need more space for one-axis rotation and far more space for two, clustering panels in small groups.<p>In my personal case I have 12+9 classic chains modules, I need more than 2x physical space to transform them with a dual-axis tracking setup. That means it's cheaper just add some fixed panels eastward and westward to catch extra power earlier and later.<p>Also in those terms: lithium storage is <i>very</i> expensive BUT for self-consumption is still the cheaper option to have electricity for more time, just arriving to a meaningful production 1/1.5h earlier and later in the day does not help much given it's added cost.<p>In costs terms: these days it's even cheaper (in TCO terms) having hot water heated by p.v. than the more efficient thermal because that cost more, have more moving parts and regular maintenance that just making an a bit bigger p.v.<p>The real issue in all cases is that to have enough power to really pay back the investment "quickly" we need much non-shadowed southward space witch can be found somewhere but far from everywhere. A similar issue is for EVs: I like the idea of charging them "for free" from solar, BUT since I normally use a vehicle during the day or I use it only sometimes or I have two or more in a round-robin scheme. Also lithium storage lifetime is an issue, on scale the production capacity and recycling are issues. Until we solve them just produce some more Wh it's meaningless...
If you know the lat/lon of the panel (they don’t move right?) and the current clock time, just point it at the current Sun position if it’s daytime? Wouldn’t that be as optimal as possible?
I manage a solar micro grid that powers 9 Holmes, a small farm, and several other structures and utilities.<p>We use intelligent power management over the fiber network to tell certain loads when to turn on or off or to change their operating parameters based on power conditions.<p>I’ve been daydreaming about building a ml based forecaster that just gives the next few hours weather outlook based on pressure, temperature, humidity, and a wide Nigel image of the sky.<p>I know it is doable because I can do it myself, and probably without any intuition about the pressure. It would automatically calibrate the model wights by feedback from the actual events vs the forecast. This would be really useful for me at least, in managing battery usage and otherwise managing the various systems that store energy like air compressors and large mass refrigeration.
This is very cool, and a neat application / exploration of RL. Am I correct in understanding that the results are only from the results of simulation? If so, it'd be cool to see this work (both a time-lapse video and a chart) in the real world, over a day or over a week!
Dyson Sphere Program - Solar Ring vs. Polar Solar<p><a href="https://www.youtube.com/watch?v=qguTFa9tj3c" rel="nofollow">https://www.youtube.com/watch?v=qguTFa9tj3c</a><p>Dyson Sphere Program · Covering Half a Planet with Solar Panels<p><a href="https://www.youtube.com/watch?v=MKxkWgknkco" rel="nofollow">https://www.youtube.com/watch?v=MKxkWgknkco</a><p>Dyson Sphere Program - Solar Panels<p><a href="https://www.youtube.com/watch?v=yO78pXYnjFA" rel="nofollow">https://www.youtube.com/watch?v=yO78pXYnjFA</a><p>Full day night cycle of solar panels | Dyson Sphere Program<p><a href="https://www.youtube.com/watch?v=gmJr4HiVCwE" rel="nofollow">https://www.youtube.com/watch?v=gmJr4HiVCwE</a>
> While primary factors driving optimal panel positioning are readily modeled (i.e., the sun’s position at each time of day), site-specific and panel-specific factors are less so. Elements like dynamic shading from nearby trees or structures, localized panel defects, drift in axis positions as systems degrade, etc. can have significant impacts on energy production.<p>Will changing the orientation of the panel really have some effect on these (besides the drift in axis position) ?
Optimisation has moved beyond the individual panel onto things like vertical bifacial panels in West/east orientation aiming to complement the fixed south facing output of other panels (and so get paid/avoid higher costs) or roof tile integrated to skip an install step and reduce transmission peaks.