VibroSens Analysis Tools

This folder contains four Python scripts for analyzing vibration data from PCE-VDL 24L 3-axis sensor.

Which script should I use?

Quick guide:

• gui_vibro.py for interactive file pick + time window + plots + metrics.
• gui_vibro_log.py for GUI analysis plus ArduPilot log overlay + G clip.
• analyze_vibro.py for batch spectra/spectrograms with automatic segmentation.
• interactive_spectrogram.py for interactive spectrograms in HTML.
• quantify_vibro.py to compare vibration metrics across tests.

Requirements

• Python 3.10+
• Packages: numpy, scipy, matplotlib
• Optional (for interactive HTML): plotly

Install dependencies:

python -m pip install numpy scipy matplotlib plotly

1) analyze_vibro.py

Batch analysis: spectra + spectrograms + automatic segmentation.

Basic usage:

python analyze_vibro.py vibro_sens_long.csv

Outputs are written to:

analysis_outputs/<file-base>/

Energy-based segmentation (adaptive threshold):

python analyze_vibro.py vibro_sens_long.csv \
  --segment-mode energy \
  --energy-k 6 \
  --min-segment-sec 10 \
  --min-gap-sec 10

Manual time window (seconds from start):

python analyze_vibro.py vibro_sens_long.csv \
  --segment-mode manual \
  --segment-start 12.5 \
  --segment-end 38.0

Key options:

• --segment-mode gap|energy|manual (default: gap)
• --segment-start / --segment-end (manual window in seconds)
• --energy-k (MAD multiplier for energy thresholding)
• --min-segment-sec (minimum flight duration)
• --min-gap-sec (minimum gap between flights)
• --energy-window-sec (RMS smoothing window)
• --clip-g (cap acceleration values to +/- this threshold in g)
• --output-dir (custom output folder)

Outputs include:

• *_segmentN_spectrum.png
• *_segmentN_spectrogram.png
• *_combined_spectrogram.png
• *_summary.txt

Example outputs:

2) interactive_spectrogram.py

Interactive spectrogram viewer. Works in headless mode via Plotly HTML.

Single axis (combined magnitude):

python interactive_spectrogram.py vibro_sens_long.csv --axis mag --plotly-out vibro_mag.html

All axes stacked:

python interactive_spectrogram.py vibro_sens_long.csv --axis all --plotly-out vibro_all_axes.html

Example (stacked axes):

Useful options:

• --axis ax|ay|az|mag|all
• --fmax 300 (limit max frequency shown)
• --max-samples 300000 (downsample for speed)
• --nperseg 2048 --noverlap 1536 (STFT windowing)

Interactive time-range selection (Matplotlib GUI):

• Press s to set range start at the cursor time.
• Press e to set range end at the cursor time.
• Press c to clear the selection.
• When both are set, peak frequencies for the selected window are printed.

Notes

• The Matplotlib GUI backend does not work in headless environments. Use --plotly-out to create an interactive HTML you can open in a browser.

3) quantify_vibro.py

Quantify vibration metrics to compare test setups.

Compare multiple files (ranked by RMS magnitude by default):

python quantify_vibro.py Test_1.csv Test_2.csv Test_3.csv

Per-file time windows (place --start/--end after each file or folder):

python quantify_vibro.py Test_1.csv --start 10 --end 60 Test_2.csv --start 5 --end 40

Sort by a different metric and save a summary CSV:

python quantify_vibro.py data\ --sort-by p95_mag --out-csv vibration_summary.csv

Cap acceleration values before metrics:

python quantify_vibro.py Test_1.csv Test_2.csv --clip-g 0.2

Available metrics for sorting:

• rms_mag (default)
• p95_mag
• peak_mag
• crest_factor
• duration_s
• samples

4) gui_vibro.py

GUI for selecting a CSV, choosing a time window, and generating spectra, spectrograms, and metrics.

Run:

python gui_vibro.py

Features:

• File picker + start/end input
• Preview magnitude vs time with hover readout
• Axis plots (spectrum + spectrogram) with hover readout
• Output images + summary metrics file

5) gui_vibro_log.py

GUI for vibro analysis with an optional ArduPilot log overlay and G clipping.

Run:

python gui_vibro_log.py

Extra features:

• Optional ArduPilot CSV/BIN overlay (PWM/RPM on secondary axis) with time offset.
• Offset controls: entry + coarse/fine +/- buttons (clamped to +/- 360s).
• Separate PWM/RPM plot window.
• Overlay auto-starts at the first PWM change away from ~1100 (idle) to make alignment easier.
• Warning message when log time does not overlap vibro time (adjust offset).
• G clip (abs) [g] to cap values above a threshold for plots and metrics.

ArduPilot .bin support:

• Requires pymavlink (python -m pip install pymavlink).

Screenshots

Existing examples:

• Combined spectrogram: media/combined_spectrogram.png
• Manual window outputs: media/combined_spectrogram_manual_60_258.png, media/segment1_spectrum_manual_60_258.png, media/segment1_spectrogram_manual_60_258.png
• Interactive stacked axes: media/interactive_all_axes.png