Monitoring, diagnosis, predictive maintenance and optimization are core challenges in the operation of technical systems. Valuable information on wear, causes of faults and optimization potentials are contained in the process data of machines and systems. However, this information often remains unused today because the quantity and complexity of the data is too high and cannot be found by humans without technical support. Therefore, we develop solutions for the easy acquisition, management, visualization and analysis of large industrial data volumes. In analysis, methods of machine learning and artificial intelligence have the potential for automated extraction of information from process data.
The Machine Learning group supports you in the analysis of existing data, develops a strategy for the introduction of data analysis methods together with you, and provides customized solutions for your challenges.
We take the first steps with you in the field of AI and enable you to roll out solutions and identify new use cases.
Our range of services
- Potential and concept studies
We check whether and which potentials lie in the data of specific machines (prediction of errors / maintenance requirements, optimization of energy efficiency, increase of cycle time / throughput / output, quality) and with which procedures the information can be made usable. We also evaluate technologies for simple yet secure data acquisition and storage and develop strategies and concepts for the integration of these technologies into the company's IT architecture.
- Design and implementation of ML integration projects in the company
Condition monitoring, surveillance, diagnosis, predictive maintenance. These are applications that add value and can be implemented with the help of machine learning. We support you in the conception of ML introduction strategies and project more extensive implementation projects.
- Cycle time analyses, bottleneck analyses
We analyze your production in terms of desired and actual cycle times and identify bottlenecks in your production. This will enable you to increase output, which will increase the overall plant efficiency.
- AI for existing installations (retrofitting)
You have old machines that should not be replaced? But you want to monitor them automatically with modern methods? According to your requirements we develop algorithms to solve your use case after extensive data recording. From statistical analyses for the extraction of KPIs, over simple monitoring methods up to deep learning methods as AI based decision system for quality monitoring or anomaly detection we accompany you from the existing machine to a modern cyberphysical system with your own AI. Thus, even old machines are made fit for Industry 4.0.
- Integration of ML processes in embedded components
You want to integrate methods of machine learning or artificial intelligence into your products? We support you in project planning and provide know-how for the embedded use of artificial intelligence. With the SmartBox, Fraunhofer IOSB-INA holds a patent for a self-configuring method for independent monitoring of production plants via network access.
- Intelligent error analysis and control
Errors in production plants often occur unexpectedly and their elimination usually requires a lot of time and expert knowledge. With the help of artificial intelligence, the causes of errors can be quickly identified, and suitable measures can be automatically initiated to minimize the effects of the error and, in the best case, even eliminate the error.
Know-how and equipment
- Analysis of large amounts of data
- CPU Rechencluster
- Parallelisation of data analyses
- GPU Cluster
- Integration of AI in existing machines
- INAsense - System for acquisition and evaluation of sensor data
- Use of ML in production
- Various demonstrators in the SmartFactoryOWL