Zum Hauptinhalt springen

Artificial Intelligence and Sustainability

Omnipresent digitalization in sustainable development results in increasing amounts of data in all relevant areas of environmental informatics, like renewable energy, environmental health, circular economy, green IT, transport, logistics, agriculture, photovoltaic, heating networks, power grids, urban ecology, nature-based solutions, building industries etc. . Applications of artificial intelligence are able to process these data as high dimensional data in very big data sets, to analyze them with goals for predictions, for recognition and interaction challenges for a more sustainable environmental protection under climate change, energy transformation and pollution conditions.

The goal of the workshop is to bundle up experts of Artificial Intelligence into a discussion on actual AI technologies which are successful in all these environmental applications and in further development on the AI-technologies.

Target group:

  • Researchers from universities and industrial companies
  • Industrial and communal partners, which are applying AI methods
  • Students, Ph-D’s and Post Docs.

Topics of the track

A Qualitative Literature Review on Machine Learning Techniques for Predictive Maintenance

The combination of increased cost pressure and increasing digitalization has created the basis for the increased use of predictive maintenance. The enormous amount of data that accumulates in the industrial environment is to be analyzed in order to prevent future failures of production capacities. In recent years, machine learning methods have emerged as a way to address these challenges. This paper shows the current state of research by using a qualitative literature review and answers the question which machine learning techniques are being researched for the use of predictive maintenance. The goal of this paper is to present an overview of the state of the art of the applied and investigated machine learning techniques in the field of predictive maintenance. The results of this study show a disproportionately high level of research activity in the manufacturing industry. It could also be shown that research interest in the public sector is underrepresented, especially in the infrastructure sector.  Furthermore, it could be shown that the focus of applied machine learning techniques can be assigned to supervised learning methods.

Identification of behavior changes in energy consumption behavior with machine learning

Within the research work for the BMBF project ENVIRON we have developed an algorithm in which behavior-relevant phases in correlation with energy consumption can be trained by a machine learning algorithm and evaluated accordingly. The focus is on phases that one or more persons pass through with regard to their energy consumption behavior, the so-called "stage model of self-regulated behavioral change" (SSBC). These phases will be detected by the algorithm in order to better analyze a long-term rebound effect in energy consumption behavior and to support the use of phase specific interventions with the goal of reducing CO2 emissions in apartments.

Recognition of different PV-module-faults based on LSTM neural network classification with the ground truth of expert labeled data from different PV-fields and modul types

In this work, photovoltaic monitoring data together with weather data are used to classify faults using an LSTM neural network. The label data for this model comes from a previous work where unsupervised models were used to detect anomalies on PV plants for an interpretation by experts.

Clustering Analysis of E-Waste Management in BRICS and G7 Countries

The management of waste electrical electronic equipment (WEEE or e-waste) has motivated the development of regulatory instruments in several countries, as well as specific management practices.
However, the formalization of processes occurs differently according to the strategies and motivations of each nation. This study aimed to analyze the similarities that can indicate the main drivers to the adoption of sustainable practices for e-waste management, using as a case study the BRICS and G7 countries. The methodology was based on multiple regression analysis and k-means clustering algorithm, respectively, considering Gross Domestic Product (GDP) and e-waste generation as indicators. The findings suggested that the economic blocs partially influence e-waste management and socioeconomic indices. Indicators such as Human Development Index (HDI) and e-waste generation have a divergent influence, while GDP uniformly influences the population and
has a significant impact. The clusters formed to show the importance of the e-waste generation potential as a determining factor, an aspect also observed in the correlation analysis. In this way, the analysis tools are complementary and reinforce the possibility of using key indicators in e-waste management.


Contributions should be written in English. The workshop language is English.


Prof. Dr. Grit Behrens, Bielefeld University of Applied Sciences

Prof. Dr.-Ing. Carsten Gips, Bielefeld University of Applied Sciences

Please contact the organizer in case you are not sure whether your working topic is of interest for this special track!

Important rules for submissions and dates can be found here.