M3-2.2 - A Data-Driven Multi-Configuration Parameter based Characterization of Eddy Current Inductive Sensors’ Signals using Machine Learning

Event

23. ITG/GMA-Fachtagung Sensoren und Messsysteme 2026
2026-06-09 - 2026-06-10
Nürnberg

Band

Vorträge

Chapter

Modellbildung und Signalverarbeitung 2

Author(s)

A. Khan, A. Hetznecker, M. Kagerer, T. Greiner - Pforzheim University, Pforzheim

Pages

456 - 462

DOI

10.5162/sensoren2026/M3-2.2

ISBN

978-3-910600-11-9

Price

free

Abstract

Eddy current based inductive sensors are widely used in industrial systems for non-contact material characterization, thickness measurement, and defect detection. However, their deployment typically requires extensive manual calibration of hardware and software parameters, resulting in time-consuming and error-prone configuration procedures. This paper presents a generalized data-driven characterization framework using machine learning, in which raw sensor signals are jointly processed with configuration parameters. A deep neural network learns nonlinear dependencies between sensor settings, electromagnetic responses, and target variables. Experimental validation using an industrial eddy current sensor demonstrates significant improvements in material identification and zinc layer thickness classification accuracy, enabling adaptive and task independent inductive sensing.