KI-MUSIK 4.0

Intelligent technologies present significant competitive advantages in manufacturing. High-performance microelectronics in combination with sensor technology and embedded software are responsible for gathering and processing process data in industrial plants. This already today facilitates a comprehensive digitisation of manufacturing processes and operating procedures in Industry 4.0.

The »KI-MUSIK4.0« (Microelectronics-based Universal Sensor Interface with Artificial Intelligence for Industry 4.0) project is dedicated to the next generation of intelligent, increasingly autonomous manufacturing systems. The aim is to make the latest electronics technologies together with artificial intelligence methods (AI) usable for industrial applications. This new quality of data processing on the spot facilitates reliable decentralised analysis and forecasting.

Innovative acoustic sensors can deliver a lot of information about machine status for the monitoring of industrial plants and machines. In the framework of the project, sensors are being developed for use in a distributed network. A special feature is that each sensor can pre-process and analyse large volumes of data by means of AI. As a consequence, each sensor unit transmits far less and much more reliable information to the base station. The quantities of data to be transmitted are considerably reduced, communication is accelerated and the user can monitor the production plant in real time. Alongside the development of innovative, energy-efficient sensor concepts, the planned research work also includes elements of machine learning.

Expected project results from real production processes point out the considerable economic potential of the microelectronic components and systems. The innovative sensor concepts are more reliable, faster and energy-saving. If the project is successful, a wide range of different applications, especially in the area of Industry 4.0, can profit from its results and findings – for example, the predictive maintenance of production plants.

• Development of distributed, intelligent, single-channel and multichannel acoustic sensor systems
• Research into AI methods in the field of machine learning for the acoustic error and status monitoring of machines and plants in individual applications and their deployment in sensor networks
• Investigation and further development of signal pre-processing methods and processes in order to achieve the following objectives:
  • Reduction of application-specific disturbance variables
  • Increasing the robustness of AI methods
  • Reduction of data volumes for resource-efficient further processing and modelling as well as for performance optimisation, e.g. for the status detection of plant and machine components
• Research into methods for the presentation – in digital form and a manner conducive to AI concepts – of expert know-how in the sense of an expert-in-the-loop approach (methods contribute to the efficient development of monitoring applications and knowledge bases and thus to improving the monitoring of industrial processes)