Master thesis: Accurate Estimation of Rooftop PV Potential using Neural Networks
Forschungszentrum Jülich
This job is no longer available.

Master thesis: Accurate Estimation of Rooftop PV Potential using Neural Networks

  • PhD / Postdoc / Thesis (From 25 to 36 months)
  • Jülich (Germany)
  • Published on August 16 2021
This job is no longer available.
Master thesis: Accurate Estimation of Rooftop PV Potential using Neural Networks

With an aim to address the climate crisis, the IEK-3 develops energy system models to analyze possible future energy systems with high shares of renewable energy in Germany and beyond. The main goal of the institute is to design cost-optimal and feasible energy systems which can aid decision-makers in politics and business.

Rooftop photovoltaic (PV) installations increase the shares of renewable energy, without requiring open areas in the region. The potential of such installations goes unexploited if a detailed study of the rooftops in the region is not undertaken. Through this master thesis, we intend to improve the estimation of the total rooftop PV potential of any globally given area. This information supports different decision-makers in the design of renewable supply systems.

Your Job:
At IEK-3, a methodology to estimate the rooftop PV potential by analyzing satellite data is implemented. We aim to improve the method further by

  • generalizing it to countries around the globe
  • further improving the accuracy by, for example:

    • detecting obstacles such as trees that cast a shadow on the PVs, thereby lowering its output
    • detecting and excluding roof areas where superstructures such as chimneys or windows are present
    • detecting and excluding roofs with existing PVs and unavailable roofs such as that of a church
Based on this, your workflow can be divided into the following steps:

  • Getting familiar with the existing methodology and codebase
  • Literature review:

    • On existing methods to detect obstacles, superstructures, building type, and existing PVs
    • On image recognition, classification, and clustering using neural networks
  • Implementation of suitable image clustering technique to classify different regions of the world based on the roof-types in the regions
  • Tuning and re-training the existing model for each identified cluster separately.
  • Implementation of a suitable image recognition and/or classification technique to identify obstacles, superstructures, building type, and existing PVs
  • Validation and comparison of the results with those of existing studies
  • Writing the thesis
Your Profile:

  • Very good academic marks in computer science, mechanical engineering, electrical engineering, mathematics, physics, or a related field of study.
  • Interest in topics and issues related to energy technology.
  • Ability to work autonomously and analytically within a project team.
  • Ideally you already have experience in modelling and programming (preferred in Python)
  • A high affinity for AI related topics.
Our Offer:

  • A pleasant working environment within a highly competent, international team in one of the most prestigious research facilities in Europe.
  • You will be supported by top-end scientific and technical infrastructure as well as close guidance by experts.
  • You will have the opportunity to work with excited researchers from various scientific fields and take part in the design of a future European energy system.
  • Your work is remunerated.
  • Depending on your performance, the small work packages can be adapted.
Ms. Shruthi Patil
Institute of Energy and Climate Research (IEK)
IEK-3: Techno-economic Systems Analysis
52425 Jülich
Telefon: +49 2461 61-6689
E-Mail: [email protected]
This job is no longer available.