Deep Learning for Power System Optimization

Project Title:
Optimization of Generator Dispatch Using Deep Learning
Project Description:
I am seeking a skilled freelancer to develop and demonstrate a deep learning-based solution for optimizing generator dispatch in a power system that integrates wind power.

The aim is to create a fast and efficient model (preferably using a feed-forward neural network) that replaces conventional iterative optimization solvers for economic load dispatch (ELD). This model should account for the uncertainty and volatility of short-term wind power, which follows a beta distribution.
Scope of Work:
Model an ELD problem that includes:
15 thermal generators with known cost functions and generation limits
100 wind turbines, each with stochastic power output based on beta distribution (1-hour and 48-hour horizons)
Generate synthetic datasets using standard optimization solvers (MATLAB)
Train, validate, and test a deep learning model to predict optimal dispatch decisions
Evaluate and compare performance with conventional optimization (accuracy and runtime)
Provide clear visualizations (e.g., convergence plots, error histograms, scatter plots)
Deliver source code and a brief report summarizing the approach and results
Requirements:
Solid background in power systems, economic load dispatch, and renewable energy modeling

Proficiency in deep learning, specifically feed-forward neural networks
Experience with MATLAB or Python (including optimization and DL libraries)
Familiarity with probability distributions, especially beta distribution modeling
Deliverables:
Source code (MATLAB or Python), well-documented
Plots/figures showing training, testing, and comparison results
Runtime and accuracy comparison with a conventional solver
Short technical report explaining the methodology, assumptions, and conclusions