Optim_SRM

A Python-based genetic algorithm system for optimizing motor parameters using OpenMotor simulations. Or simply a program for SRM design optimization.

This project uses evolutionary strategies (selection, crossover, mutation) to find optimal motor designs. The simulation is driven by .ric input files and external Python simulation scripts.

---

🧠 Features

• Genetic Algorithm with customizable selection, mutation, and evolution thresholds
• Compatible with .ric or .json inputs
• Automatically creates timestamped directories for each run
• Outputs .json and .ric files for the best results
• Command-line interface using argparse
• Supports seeding for reproducibility

---

🛠 Requirements

• Python 3.8+
• openMotor simulation environment and .venv path in it should be set up following the openMotor installation instructions

You may install requirements like:

pip install -r requirements.txt

🚦 Usage

Run the optimizer via command line:

(.venv) python gen.py [OPTIONS]

Available CLI Arguments:

Argument	Description	Default
--script-path	Path to the simulation script to run	openMotor/main.py
--input-file	Input motor data file (.json or .ric)	data.json
--limits	Limit type: 'Default', 'RIClike', or a custom JSON path	'Default'
--n-populations	Number of individuals per generation	10
--evo-threshold	Stop if score improves less than this threshold	0
--max-same-results	Stop if best score doesn't improve for this many generations	5
--seed	Random seed (generated if not provided)	None
--naive-tol	Tolerance when interpreting .ric inputs into parameter limits	12

Example Run

Run the script with a .ric input file and custom settings:

(.venv) python gen.py\
  --input-file motor.ric \
  --script-path openMotor/main.py \
  --n-populations 20 \
  --evo-threshold 5 \
  --max-same-results 3 \
  --seed 12345 \
  --limits Default

If you want to use my example .json input file (or your manually crafted), you can use that instead:

(.venv) python gen.py --input-file optimized_design.json

🗃 Output Structure

Each execution creates a timestamped folder like:

2025-02-25_16-03-12/
├── inicializace_<SEED>/
│   ├── iter_0_score_10.23.json
│   ├── iter_1_score_9.56.json
│   └── ...
├── postup_<SEED>/
│   ├── iter_1_score_8.88.json
│   ├── iter_2_score_7.34.json
│   └── ...
├── result_<SEED>.json
└── GEN_RESULT_<SEED>.ric

📚 Project Structure

.
├── gen.py/                    # Main CLI runner
│   ├── Simulation             # Handles running the simulation script, collecting output and output evaluation
│   └── Population             # Performs initialization, selection, mutation and breeing of population
├── data.py/
│   ├── DataGenerator          # Handles .ric → JSON conversions
│   └── RICFileHandler         # Reads/writes .ric files
├── README.md
├── openMotor                  # Simulation interface
├── transform                  # Transform function that rescales and shifts the simulated curve
└── data.json                  # Example input data

📈 Scoring System

Each individual's "fitness" (score) is calculated based on how close it is to given parameter limits:

• Inside bounds = no penalty

• Outside bounds = scaled penalty based on a factor

• The total score = sum of all parameter scores

• The algorithm tries to minimize this total score over generations.

👤 Author

Developed by Hana Josifkova.