Reinforcement Learning

Electricity prices have risen significantly year on year and reducing energy use in homes can save money, improve energy security and reduce pollution from non-renewable energy sources. Whether to lower the monthly electricity bills or be concerned about the home's carbon footprint, reducing energy is helpful. The best way to start saving on electricity costs is to get smart with how electricity is being used. The goal of this paper is to find an efficient approach to using electricity using machine learning algorithms. To achieve that, this thesis will apply q-learning, DQN with Replay Memory, and Double DQN with Replay Memory of Reinforcement Learning in python. The agent will interact with the Gym environment implemented from data given by Nova Smart company, achieving rewards upon reaching the goal or penalties based on the power price, time of the day, and comfort zone. Numerous studies have been conducted on this subject recently and there has been a lot of research. This work will demonstrate the behavior of the algorithms to meet the main criteria of trajectory design as an alternative solutio