Detection of early symptoms of equipment malfunction could prove to be a major cost saver within the industrial sector. A sensor network being developed by Professor Andreas Schütze and his research team at Saarland University will diagnose developing failures in industrial machinery based on the systemic sensory response to changes in vibrational frequency and temperature.
The Saarland University and Center for Mechatronics and Automation Technology (ZeMA) team, along with the German Research Center for Artificial Intelligence (DFKI) and the HYDAC group, presented their system using a hydraulic test bench at the Hannover Fair in April. They recorded continuous feedback from sensors at different coordinates on the machine to correlate network response with specific component or system malfunctions and failure modes. They also use feedback from standard sensors already on the equipment.
The team plans to use statistical methods, mathematical models, and raw data analysis to create algorithms for use with a multitude of different machines. “The aim is to develop the system so that it can be trained to work with different types of machine and plant equipment, and can be adapted and customized to meet their specific requirements,” says Schütze.
The engineers can then “teach” the network to generate the warning that fits with its response. This means that operators can replace parts when they need replacement, instead of relying on lifetime warrantees, while also cutting down on maintenance fees and emergency system shutdowns for repair. The system also has self-checking abilities to build in more network reliability.