Skip to content

v0.2-beta Measurements; OCR; Stability fixes

Pre-release
Pre-release
Compare
Choose a tag to compare
@mj-xmr mj-xmr released this 22 Jul 14:36
· 106 commits to master since this release
v0.2-beta
3f4f1f0
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Intro

This pre-release is one step before being able to automate the act of switching the mining on and off. It deals with supplying all the necessary measurements, yet still without the "Home Assistant" hardware, but rather the bare metal ones. It automates the part of acquiring the initial battery state - the only part that had to be supplied by hand to perform the simulation. Moreover, by adding a few more parameters to the battery.json, described in the configuration documentation, the required conversions from voltage to charge percentage are being performed.

Major changes

Charging and discharging voltage modelling

Thanks to the voltage modelling, we can now translate the voltage readouts to battery charge in percent, yielding a linear calculation, ultimately being able to translate voltage to Ah (Ampere-hours): a unit of energy capacity, that's much simpler for calculations.
As described in the main documentation, the voltage input can be triggered with:

./soloptxmr.py --battery-charge-v 12.3 # Set the voltage of your battery to be converted to its current charge

OCR

Thanks to an incorporation of GasPumpOCR project, it's possible to reuse the existing examples of OCR modules or to write your individual ones. The right parameters of such a module can be learned with an interactive script called playground.py of the GasPumpOCR project (example). See the new documentation. The OCR module can be generalized to just recognize certain rectangles, that could represent the battery charge in % already, rather than the raw voltages, like here . There's an example that does it as well.
Using the OCR module is done via:

./soloptxmr.py --battery-charge-ocr # Use image recognition to read the current battery voltage

Image capture

Closely related to OCR (and documented there) is a similarly interfaced image capture system, allowing you to write your own image capture models.

Temperature readouts

These will come very handy whenever there's a suspicion, that a given rig is overheating. In such an event the controlling PC will reduce the mining effort of the mining rig to let it cool down.

Optimization algorithm stability fixes

I noticed, that the algorithm behaved in an unstable way for a few reasons not worth mentioning. This part is simply being left in a working ad-hoc state for now, while I focus on the integration of many individual systems, like the ones mentioned above. There will come a time soon, when I'll focus on making the optimization algorithm work much better.

Full changes

What's Changed

New Contributors

Full Changelog: v0.1-beta...v0.2-beta

Contributors

dependabot and mj-xmr
Assets 2
Loading