This project is co-financed by the European Union from the european fund for regional development. Operationelles Programm Mecklenburg-Vorpommern 2014 – 2020 – Investitionen in Wachstum und Beschäftigung .
Continuous condition monitoring is indispensable for condition-oriented maintenance of large-scale power engines in operation, such as ship engines. Faults in machines usually manifest themselves through a change in the noise signature long before a critical condition occurs. Measurement with acoustic sensors already makes it possible to test individual components. In this project, the individual sensor system, which has to be developed, is to be expanded into a sensor network, enabling continuous monitoring of the entire life cycle. Efficient use in the sense of a "condition monitoring system" requires Industry 4.0 approaches: digitalisation, networking and "machine learning". Acoustic monitoring of the entire life cycle thus forms the basis for the development of solutions for the megatrend of predictive maintenance.
The aim of the project is to develop a new type of sensor platform with network capability and adapted analysis and sound localisation procedures. The distributed measurement of sound events opens up unique possibilities for the comprehensive monitoring of large units with just a few sensors, expecially through the linking of sensors and sound localisation. In addition to the development of the sensor network and the localisation of the sound events, methods for the optimal placement of the individual sensor systems are being investigated.
This is a collaborative project between ds automation gmbh and Fraunhofer Institute for Large Structures in Production Engineering (IGP).
This project is co-financed by Federal Ministry for Economic Affairs and Climate Action.
The application scenario for the sound sensor is intended for use in the precise monitoring of filling plants, conveying equipment and dosing systems.
Machine learning (ML) methods are used to extract the features from the audio signals that are essential for detecting a fault in the process or damage to the system.
Process faults affect the quality of the products and machine damage can lead to longer and more expensive downtimes if not detected in time. To ensure consistent quality and avoid production downtime, defect detection is essential. The features obtained are evaluated by an artificial neural network. Feature extraction and the neural network are optimised so that they can be run embedded on a microcontroller. The manual effort for creating the sensor configuration (testing, training and configuring) is to be replaced by the KNN and thus automated. Among other things, this should reduce costs during commissioning and operation.
The aim is to develop a new sensor platform with an integrated artificial neural network.
This is a collaborative project between ds automation gmbh and Fraunhofer Institute for Microelectronic Circuits and Systems (IMS).