PSO Optimization for Self-Potential Data

For an in-depth explanation of the methods and results, check out my Medium blog: Optimizing Self-Potential Models with Particle Swarm Optimization: A Geophysical Approach

Overview

This project involves optimizing the parameters of a self-potential model using Particle Swarm Optimization (PSO). The self-potential method is used in geophysics to detect anomalies in subsurface materials. The goal of this project is to find the optimal parameters that best fit a self-potential profile using PSO.

Project Structure

1. analyticalSig.py: Contains the main script for running the PSO optimization, generating plots, and creating a GIF of particle movements.

2. potential.py: Defines the potential functions used in the optimization:

   • PotentialSphere: Calculates the potential due to a sphere.
   • PotentialCylinder: Calculates the potential due to a cylinder.
   • AnalyticalSignal: Calculates the analytical signal.

3. logs/pso_log.log: Log file that records the details of the PSO optimization process.

4. images/: Directory where results are saved, including:

   • Error PSO.png: Plot of error vs. number of iterations.
   • SP_Profile.png: Plot of self-potential profile.
   • particle_movement_PSO.gif: GIF showing particle movements during optimization.

Installation

Ensure you have the following Python packages installed:

• numpy
• matplotlib
• tqdm
• imageio

You can install these packages using pip:

pip install numpy matplotlib tqdm imageio

Usage

1. Run the Optimization: Execute the analyticalSig.py script to start the PSO optimization process. This script will:
• Initialize the PSO parameters.
• Optimize the parameters using PSO.
• Save the results and generate plots and GIFs.

pip install numpy matplotlib tqdm imageio

2. View Results: After running the script, the following files will be generated in the images/ directory:
   • Error PSO.png: Shows the error vs. the number of iterations.
   • SP_Profile.png: Compares the actual self-potential profile with the optimized profile.
   • particle_movement_PSO.gif: Visualizes the particle movements during the optimization process.

Description

PSO Optimization

• Initialization: Particles are initialized randomly within a specified range.
• Velocity Update: Particles’ velocities are updated based on their personal best and the global best positions.
• Position Update: Particles’ positions are updated based on their velocities.
• Error Calculation: The error is calculated as the root mean square difference between the predicted and actual self-potential values.
• Stopping Criteria: The optimization stops when the error falls below a predefined threshold.

Potential Functions

• PotentialSphere: Calculates the self-potential due to a sphere.
• PotentialCylinder: Calculates the self-potential due to a cylinder.
• AnalyticalSignal: Calculates the analytical signal for comparison.

Logging

The optimization process is logged in logs/pso_log.log. The log file contains information about the optimization runs, including:

• The number of iterations.
• Parameters used in each run.
• Error values.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

The self-potential model and optimization techniques used in this project are based on established geophysical methods and optimization algorithms.

Contact

For any questions or feedback, please reach out to:

Raj Sahu
Email: sahuraj457@gmail.com