Optimization of Factory Energy Costs through Solar Energy Forecasting and Machine Scheduling Using Mixed-Integer Linear Programming

This study utilizes machine learning and optimization techniques to lower energy costs in factory operations. By incorporating weather forecast data and machine learning techniques, we aim to accurately predict solar energy production. Various machine learning models, including XGBoost, Random Forest, Support Vector Regression (SVR), Ridge Regression, and Long Short-Term Memory (LSTM) networks, were employed. Among these, XGBoost emerged as the most effective model for solar energy forecasting. Utilizing the solar energy production predicted by the models, a Mixed-Integer Linear Programming (MILP) optimization algorithm is used to determine the optimal scheduling for Sand Milling Machines. The model strategically schedules machine operations during intervals when solar energy is underutilized and during off-peak periods to minimize energy costs. This approach also ensures that the sand milling process meets critical constraints by maintaining sand levels within desired minimum and maximum thresholds. The results underscore the potential for significant cost savings and enhanced energy efficiency in factory operations through the integration of machine learning predictions and optimization techniques.