Master Thesis: Quality Enhancement of OpenStreetMap Data Using Machine Learning for Data Driven Modeling

Jülich Research Centre (FZJ)

Jülich, Germany

Work group:

IEK-3 – Elektrochemische Verfahrenstechnik



Area of research:

Diploma & Master Thesis



Starting date:

1569335620



Contract time limit:

1569335620



Job description:

Start of work: as soon as possible/by arrangement


Framework:The need to reduce greenhouse gas emissions lead to increasing penetration of renewable energy sources and market penetration of efficient end-use technologies like heat pumps or battery electric vehicles to boost up renewable utilization. The new electricity consumers, as well as the renewable distributed generations, bring new challenges to existing distribution grids. In this context, non-availability of real distribution networks motivate towards the development of synthetic distribution grid networks and models to access the distribution grid challenges.


The master thesis will focus on improving the quality of OpenStreetMap (OSM) data using machine learning algorithms specifically for data-driven modeling of distribution grids. OpenStreetMap is a free editable map built by crowd community. This crowdsourced data is available under the Open Database License. Moreover, OSM data quality is improving day by day but for some cases for detailed energy-related information, it is too low. The main focus of this thesis is to enhance the quality of tags concerning the building types in OSM for usage in the synthetic distribution grids and in the distribution grid model. Based on a literature survey typical building types and their characteristics should be identified. With the identified characteristics and existing tag information from OSM and other sources, the prediction of tags by developing a prediction model using machine learning algorithm is goal an important work package. The overall objective of the model is to provide prediction of the tags by comparing different machine learning algorithms for accuracy and choosing the best model approach.


Project Description:



  • Literature research on building types and their characteristics in Germany and neighbor countries

  • Literature research on machine learning algorithms

  • Data research on different sources for training the algorithms

  • Develop the prediction model

  • Compare different machine learning algorithms for accuracy

  • Applying the model for selected countries to improve data quality of building types in OSM

Requirements:

  • Excellent academic marks in Mathematics, electrical engineering, energy engineering, physics, computer science, or related fields of the study
  • Knowledge in SQL, Python, PostGIS, PostgreSQL
  • Familiarity with Machine learning
  • Excellent communication skills in English. German is a plus

Our range:

  • A highly motivated working group in one of the largest research institutions in Europe
  • An excellent scientific and technical infrastructure
  • The possibility of actively shaping the energy system of the future
  • Intensive support of the work on site

We are looking forward to receiving your application!


Please apply via recruiter’s website.

Quote Reference: 2019M-096

Favorite