Towards Battery-Free Industrial IoT : Evaluating Energy Harvesting Methods on Industrial CNC Machines

Sammanfattning

Predictive maintenance via the Industrial IoT (IIoT) entails high costs due to manual battery replacements. This thesis therefore evaluates three energy harvesting methods on a CNC machine: piezoelectric (vibrations), magnetic induction (power cables), and radio frequency (2.4 GHz). Theoretical and empirical evaluations show that no configuration delivered sufficient continuous power for a reliable cold-start of a sensor node. Magnetic induction using a split-core current transformer showed the greatest potential through non-invasive installation, but was limited by inefficient rectification at low voltages. Piezoelectric harvesting yielded only 100--110~mV peak-to-peak during continuous spindle operation, which failed to charge capacitors, whereas transient axis movements generated voltage peaks of up to 9~V. Radio frequency energy yielded at most 4.5~nW, quickly became indistinguishable from background noise, and remained far below the threshold of commercial power management integrated circuits. In conclusion, magnetic induction is the most promising method forward, provided the rectification stage is optimized.