Digital 3D Wideband Vibration Sensors for Improved Machine Monitoring Through Machine Learning

Business Unit »Intelligent Sensor and Actuator Systems«

Condition monitoring based on vibration analysis using the example of a pump.
© Fraunhofer ENAS
Condition monitoring based on vibration analysis using the example of a pump.

 In the project diVIBES a system for predictive condition monitoring based on the detection of vibrations is developed. Innovations along the entire information chain will make a significant contribution to the improvement of predictive condition monitoring and process optimization. This includes new approaches in signal acquisition, processing, networking and the use of machine learning (ML) methods. Due to its high sensory information content, vibration measurement in particular offers enormous opportunities for increasing the automation, digitization and autonomy of industrial plants within the general technology trend Industry 4.0.

The figure shows the cooperation of the partners involved and their competencies using the example of the demonstrator for monitoring pumps. There is a multitude of causes for unacceptable vibrations. These can occur at an early stage or after a long period of operation. In either case, they lead to a drastic reduction of the remaining service life. Desirable features are the detection of the vibration, an intelligent extraction of information from the detected signal and a learning algorithm that initiates the correct measures. In this way, the safety and efficiency of processes can be increased.

Link to the project profile of the BMBF: