In collaboration with a machine maker, ConSenses enhanced a production machine with predictive maintenance functionalities. In this manufacturing process, axial sub features on hollow or full shafts are produced while the work piece is formed during several recursive forward strokes. Characteristic process sequences and a representative force path are shown right.
For a safe process monitoring, several measurement layers were tested. The result of this study was, that the most speaking data could be measured at the work piece holder, directly at the process. The usage of four PiezoBolts allowed the recognition of eccentric loads which can occur due to one sided wear.
The measurement location on height of the process shows clear reactions on wear. The picture besides shows the force paths (mid/downside). The machine force sensorics is way less sensitive to changes at the tool (mid/upside). For the design of predictive maintenance processes, the edge device calculated a set of characteristic values for every part. The cloud on the right side of the picture represents one of these KPIs of one sensor. The max value of this KPI went smaller, before tolerance problems occurred within the production. The measured data is used in systems of artificial intelligence or in this case classical threshold value systems in order to make decisions in the direction of predictive maintenance.
This Article contains details (German).