Signal-Quality-Assessment

This repository contains Python scripts to assess the quality of time-series signals. The main steps include:

1. Reading raw data from text files.
2. Segmenting signals into windows.
3. Transforming them to the frequency domain (FFT).
4. Computing quality metrics (Spectral Entropy, etc.).
5. Visualizing the mean spectrum with uncertainty (standard deviation).
6. Saving metrics into timestamped .yml files.

Table of Contents

• Directory Structure
• Installation
• Usage
• Example Commands
• Details

Directory Structure

.
├── .gitignore
├── data/
│   ├── 0702 без перекосов фаза A.txt
│   └── data-sample_motor-operating-at-100%-load.txt
├── environment.yml
├── pyproject.toml
├── res/
│   ├── result.yml
│   └── segment_with_uncertainty.png
├── setup.sh
├── src/
│   ├── __init__.py
│   ├── data_preprocessing.py
│   ├── quality_assessment.py
│   ├── utils.py
│   └── vis.py
└── main.py
└── README.md

• data/: Contains example .txt signals.
• res/: Output folder for results, plots, and metrics in .yml format.
• src/: All source code modules.
• main.py: Command-line interface script for running signal quality assessment.
• environment.yml / pyproject.toml: Dependencies configuration.
• setup.sh: Setup script.

Installation

1. Clone the Repository

git clone https://github.com/xndrleib/Signal-Quality-Assessment.git
cd Signal-Quality-Assessment

2. Set Up the Conda Environment

bash setup.sh
conda activate py311_sqa

3. Install the Project in Editable Mode

Install the package using the pyproject.toml setup:

pip install -e .

Usage

Command-Line Interface

Run the CLI script main.py to process either a single file or an entire folder of signals.

python main.py --input <path_to_file_or_folder> [OPTIONS]

Options:

• --res_dir: Output directory for results (default: "res").
• --window_length: Number of samples in each segment (default: 10000).
• --step: Step size for windowing (default: 20).
• --f_sampling: Sampling frequency (default: 10000.0).
• --db: Flag to convert FFT results to dB scale (default: True).
• --cutoff_freq: Frequency cutoff for filtering (default: 250.0).

Example Commands

1. Process a single file:

   python main.py \
       --input data/data-sample_motor-operating-at-100%-load.txt \
       --res_dir res \
       --window_length 10000 \
       --step 20 \
       --f_sampling 10000 \
       --db True \
       --cutoff_freq 250

2. Process all .txt files in a folder:

   python main.py --input data/

After completion, the script will generate:

• Plot of Average Spectrum with Uncertainty in res/spectrum_with_uncertainty_data-sample_motor-operating-at-100%-load_20241227_101523.png.
• Result metrics in .yml files (e.g., res/data-sample_motor-operating-at-100%-load_20241227_101523.yml).

Details

• Signal Reading: Implemented in src/utils.py.
• Segmentation & FFT: Implemented in src/data_preprocessing.py.
• Quality Metrics: Spectral Entropy, etc. in src/quality_assessment.py.
• Visualization: Plot functions in src/vis.py.