|Theme||Production planning, Artificial Intelligence|
|Project title||Enabling SMEs to exploit the potential of machine learning in production and developing a deployment strategy (ML-Ready)|
|Project duration||01.03.2022 – 29.02.2024|
How can small and medium-sized enterprises (SMEs) in mechanical engineering be enabled to implement and use the potentials of Machine Learning to achieve an improvement in resource efficiency in production? This is what we are investigating in the ML-Ready research project together with IPRI.
Machine Learning is already being applied in many areas of everyday life, such as transportation or healthcare. The foundation for a successful application of machine learning is data and its availability as well as sufficient computing power to evaluate it. Nevertheless, Machine Learning has so far hardly been used to increase resource efficiency in production.
In the corporate environment, machine learning is mostly associated with predictive maintenance. In production, however, there are other versatile potentials for increasing resource efficiency:
- In production planning and control, Machine Learning can be used, for example, in order control for sequence determination (setup optimization, reduction of energy consumption or energy costs), for load leveling or for resource and capacity planning
- The application of machine learning for process optimization is expected to increase the adaptability of processes to changing conditions, which can stabilize product quality. For example, fault diagnostics can be performed or condition monitoring can be applied.
- In quality management, machine learning-based models can be used to monitor or predict product quality based on process data. In this way, measures such as random sample checks can be reduced.
However, when it comes to implementation, companies encounter hurdles such as lack of know-how, lack of proof of added value or lack of technical infrastructure. In the ML-Ready project, we want to support SMEs in particular in overcoming these hurdles in the future and using machine learning.