Fraunhofer Institute of Optronics, System Technologies and Image Exploitation IOSB
Fraunhofer Institute of Optronics, System Technologies and Image Exploitation IOSB

Data Analysis

AI-based data analysis enables correlations in data to be identified automatically, thereby supporting decision-making. The possible applications of AI-based data analysis in an industrial environment include:

  • Anomaly detection,
  • predictive maintenance, and
  • quality assurance.

Both time series data and image data in the form of camera recordings can be processed. We use patented solutions, state-of-the-art deep learning technologies, and innovative methods such as physics-informed neural networks and data augmentation approaches to address customer-specific use cases.

The AI methods are often implemented in close cooperation with the development of intelligent sensor systems. The goal is to realize real-time solutions in the context of embedded systems. By combining advanced sensor technology and AI-supported analyses, companies can react quickly to changes in their production processes and make more efficient decisions.

Projects

Datenfabrik.NRW

AI solutions for production and logistics

At Fraunhofer IOSB-INA, an AI-supported anomaly detection system was developed in collaboration with Schmitz Cargobull that uses images to check incoming material in production for damage or other defects. Thanks to the support provided by AI, specialists are relieved of some of their workload and significantly smaller defects are detected. A second use case is the optimal distribution of a large pool of orders across production days, freeing up additional production capacity and enabling individual customer requirements to be taken into account even better.

Find out more

Municipal climate dashboard

Bürgerwolke Soest

The project involved setting up a comprehensive sensor network in the urban area of Soest using low-cost sensors. It is capable of recording and providing a wide range of climate parameters in real time for use in planning leisure activities, agricultural decisions, and urban development. The quality of the measured values is optimized using comparative measurements with sensors from the German Weather Service and based on artificial intelligence and machine learning methods developed by Fraunhofer IOSB-INA. A special feature is the involvement of citizens – half of the 100 sensors were installed on private property.

Bürgerwolke Dashboard

Complementary ML methods

Adaptive plant monitoring with COMETH

Preparing large, unlabeled data sets for intelligent algorithms is particularly time-consuming. With changing conditions and often insufficient data, the question arises of how to avoid failures and detect anomalies at an early stage. The Fraunhofer-patented COMETH solution relies on a combination of complementary ML methods and a feedback mechanism to enable efficient and adaptive anomaly detection.

Find out more