AI-Based Heat Demand Forecasting

This repository contains the source code developed for the master’s thesis “AI-Based Heat Demand Forecasting” at Technische Hochschule Würzburg-Schweinfurt (THWS). The project integrates multi-horizon heat demand forecasting with reinforcement learning–based control optimization to improve operational efficiency, reliability, and cost-effectiveness in district heating systems. The upload includes preprocessing workflows, baseline forecasting models (MLR, SVR, FFNN, LSTM), and advanced control modules (SAC, TD3, and hybrid RL–MPC), along with all datasets, results, and utilities required for full reproducibility.