Those who want to survive in competition must have their costs under control and increase the productivity of their company. Safe operating processes through reliably working plants are basic prerequisites for this. Especially in the case of individual technical systems for the realisation of complex tasks, one must therefore be able to rely on the maintenance measures and methods. But what can be done when this reliance is no longer given?
Our customer, a medium-sized wood processor, reached the limits of all current predictive maintenance measures due to the special circumstances of his industry.
The limits of vibration analysis
Especially in the wood industry, common methods such as vibration measurements and vibration analyses for predicting bearing and motor failures fail. The existing interfering vibrations caused by the transport and passage of tree-trunks weighing several tons are so high that the relevant signals cannot be filtered out. Furthermore, measurements have to be taken during operation under constant load changes, and the common methods of "threshold crossing" are not up to this complexity.
The result: the results of the common methods lose their significance due to the problems just described.
When ultrasound meets artificial intelligence
In order to nevertheless guarantee our customer a reliable solution for his plant maintenance, ultrasound measurements were taken once a month at the bearing points of the affected plants as a first step. Subsequently, these measurements are broken down into 400 different statistical factors and then evaluated by artificial intelligence. The results are available in about 2 to 3 days.
With the help of the artificial intelligence trained by us and the input of the ultrasonic measurements, we were thus able to get much closer to the actual failure time and avoid unnecessarily early bearing changes and unplanned failures.
Author: Dipl. Ing. Mag. Markus Loinig, Senzoro GmbH