Overview

• This thesis explores the development and evaluation of AI-driven smart control strategies for residential energy systems integrating solar panels, battery storage, and the electrical grid.
• Focus areas include optimizing energy costs, maximizing solar self-consumption, and extending battery lifespan using advanced AI techniques.

Technologies & Techniques

• Primary technologies: Python, TensorFlow, PyTorch, Stable-Baselines3.
• Core techniques: Neural Networks, Reinforcement Learning, Rule-based control, and Optimization; rule-based and optimization-based methods will be used as benchmark baselines.

Objectives & Research Tasks

• Design and implement AI-driven control strategies (including RL-based controllers) to manage PV generation, battery charging/discharging, and grid exchange to minimize cost and maximize self-consumption.
• Compare proposed AI approaches against rule-based and optimization-based baselines; evaluate on metrics such as cost savings, self-consumption ratio, and battery degradation/lifetime impact.

Expected Deliverables

• A working EMS prototype (simulation and/or real-data experiments) implementing the proposed AI strategies and baseline methods.
• Experimental evaluation reports, codebase (Python) using TensorFlow/PyTorch and Stable-Baselines3, and a written thesis documenting methodology, experiments, and results.

Candidate Profile & Skills

• Good programming skills in Python and familiarity with machine learning frameworks (TensorFlow or PyTorch); experience with RL libraries (Stable-Baselines3) is a plus.
• Background in control systems, energy systems, or optimization is desirable; ability to work with simulation environments and time-series energy data.

Logistics & Application

• Location: Sfax. Duration: 6 months (remunerated).
• To apply, send your CV and a brief motivation letter to career@habemus.com specifying the internship topic number in the email subject.