Smart Home Energy Management App with PV Prediction

Co-Supervised by: Fereshteh Jafari

If you are interested in this topic or have further questions, do not hesitate to contact fereshteh.jafari@unibe.ch.

Background / Context

As residential solar installations and home battery systems become increasingly common, homeowners face complex decisions about when to use appliances, charge electric vehicles, or store energy in batteries. Without proper tools, these decisions are often made intuitively, leading to suboptimal energy usage and increased costs. Smart home energy management systems can help optimize these decisions by predicting solar generation and providing actionable recommendations. However, existing solutions are often too complex for average homeowners or lack integration between prediction models and user-friendly interfaces. There is a need for intuitive applications that combine accurate PV prediction with practical energy management guidance.

Research Question(s) / Goals

The goal is to propose an energy management method and develop a user-friendly application that enables homeowners to optimize their energy usage through:

  • Integration of PV power prediction with household energy management
  • Simple and intuitive user interface for energy monitoring and control
  • Automated recommendations for optimal appliance scheduling
  • Battery management strategies based on predicted solar generation
  • Cost-benefit analysis and savings visualization for users

Approach / Methods

The student will use an offline trained neural network for short-term PV power prediction.

  • Propose an energy management algorithm
  • Design and implement a web or mobile application using modern frameworks (React, Flutter, or similar)
  • Integrate PV prediction models and household load profiles
  • Develop simple optimization algorithms for scheduling flexible loads (washing machines, EV charging, etc.)
  • Create intuitive visualizations for energy flows, predictions, and cost savings
  • Implement user preference learning to personalize recommendations
  • Conduct usability testing with potential users to refine the interface
  • Evaluate the system using presimulated household energy consumption

Expected Contributions / Outcomes

A functional application prototype that generates control commands for:

  • Switching steerable loads on/off based on available PV generation
  • Adapting electric vehicle charging power levels according to solar production
  • Managing battery charging and discharging cycles for optimal energy storage
  • Analysis of the potential increase in PV self-consumption achieved through the implemented control strategies
  • Economic assessment demonstrating the cost-saving potential of the energy management system
  • Complete documentation of system architecture and practical deployment considerations

Required Skills / Prerequisites

  • Web development skills (HTML, CSS, JavaScript) or mobile app development (Flutter, React Native)
  • Python for backend development and API creation (Flask, FastAPI)
  • Basic understanding of optimization algorithms and energy systems
  • User interface design principles and experience with design tools
  • Database management for storing user data and energy profiles

Possible Extensions

  • Integration with smart home devices and IoT sensors
  • Machine learning for user behavior prediction
  • Grid pricing integration
  • Integration with home automation systems

Further Reading / Starting Literature

  • Luthander, R., et al. (2015). “Photovoltaic self-consumption in buildings: A review.” Applied Energy, 142, 80-94.
  • Nielsen, J. (2020). “10 Usability Heuristics for User Interface Design.” Nielsen Norman Group.
  • Wu, X., Tang, Z., Stroe, D. I., & Kerekes, T. (2022). Overview and comparative study of energy management strategies for residential PV systems with battery storage. Batteries, 8(12), 279.
  • www.enjoyelec.net, www.senkrondigital.com, www.scalosoft.com.