Forecast-informed recommender system for user actions in a photovoltaic system

  • Christian Miguel Botia Gonzalez

Student thesis: Master's Thesis

Abstract

Fronius, a global company renowned for producing inverters, which are a vital part of Photovoltaic (PV) systems, utilizes its Solar.web platform to monitor PVs and connected devices. This platform offers detailed information on system parameters, historical data, and visualizations through reports and graphs. It also allows for configuration settings and simulations for system modifications, such as adding a battery. At present, Solar.web does not assist in identifying the best times to use household appliances, such as washing machines or dishwashers, even when sufficient PV energy is available, which would otherwise be fed into the grid at a low rate.
This work focuses on creating a system that recommends the optimal times for starting household facilities, such as washing machines, dishwashers, etc. The recommendations are based on predictions of solar energy production and the historical load from solarWeb data. The system will use data from existing Application Programming Interfaces (APIs), including real-time system values, PV production forecasts, load forecasts, and possibly weather forecasts.
Subsequently, advanced machine learning techniques will be incorporated. These methods will calculate the best actions to maximize rewards, defined by energy savings and efficient use of PV energy.
By offering practical recommendations for when to use appliances, this system aims to help users make smarter energy choices, improving the efficiency of PV energy use and saving money. This research will demonstrate how artificial intelligence can enhance smart home energy management.
Date of AwardSept 2024
Original languageEnglish
SupervisorKathrin Maria Kefer (Supervisor)

Cite this

'