Raw data from Carina is stored in Data/<Test ID>/raw with the filenames data.log and events.log. The script in src/main.py first parses these into CSVs for quicker re-run times. These CSVs are stored in /cache/<Test ID>. The script stored the rest of its output in Data/<Test ID>/compiled under six files.
- Three correspond to different windows of sensor data, three correspond to different windows of event data.
- The windows are user-selectable with the
LOWERandUPPERvariables, as well as theLOWER_FandUPPER_Fvariables. The approach I've been using is to select the full test data range (from the start of fill to a few seconds after the flow) with the first window, then to select just the flow with the second window. The other output file is a copy of the cached.
Each folder in Data/<Test ID> contains a python script analysis.py which automates the creation of graphs, and some data analysis/smoothing.
The file plot_them_all.py is a generalized script that plots all the flow data from each test in a single graph.
TODO:
- Make graph generation more user-friendly than copy-pasting and editing the
analysis.pyfiles. The best solution might be to keep a majority of the code in a shared library, and then instantiate that library with per-test specifics in each individual test folder. For instance: you may wish to write a generalized function that can rename a sensor/actuator, or one that can set the timestamps to reference a specific event. Consider doing something similar for the graphing scripts. - Structure things better, make the code more dev-friendly.
- Make the code more generalizeable. Parsing should work fine, but there's a lot of potential improvement for analysis/plotting.
- Consider better ways to do range-selection. Maybe a CLI instead of setting global variables. For instance, you could have the user specify the name of a mass sensor via terminal, which is then graphed after the initial parsing. This makes finding the range(s) of interest more user-friendly than opening it in excel.