Enhanced Hybrid Neuro-Symbolic AI for External Magnetic Interference Classification in Magnetostrictive Position Sensors
Aimal Khan, Tobias König, Alexander Hetznecker, Thomas Greiner · IEEE Sensors Journal · 2025
The growing use of sensors in both industrial and commercial settings demands sensors that are flexible, precise, and reliable. To meet these requirements, especially complex sensors should be able to detect anomalies and defects in both their environment and themselves and respond appropriately. Important examples of complex sensors are magnetostrictive position sensors (MPSs), which are used for high-precision distance and velocity measurements. These sensors work on the basis of time-of-flight (ToF) calculation for a structure-borne torsional wave generated through the interaction of an excitation pulse and a marker magnetic field within the sensor systems. The accuracy of these sensors is affected by external magnetic interference (EMI) which can interact with the structure-borne sound wave they generate. This study presents a novel solution using enhanced hybrid neuro-symbolic AI to classify the intensity of EMI directly from the received distance signal without the help of an additional sensor. This work is of significance for measurement in industrial environments where multiple sources of EMI can be present.