We propose a fast method to characterize inductive position sensors with zero physical prototypes. The technique is based on an in-house developed electromagnetic simulation tool which shows three orders of magnitude improvement with respect to the most widely used commercial software. This simulation software is used for producing synthetic data for training a machine learning surrogate model based on a neural network. In this way, the characterization of a sensor takes just milliseconds. This opens the possibility of devising a Design Support System (DesSS), a software that guide the user in sensor design for the best possible outcome, thereby saving development time.
Real-Time Design and Characterization of Inductive Position Sensors Through AI-Driven DesSS
Campagna F.;Trevisan F.;Specogna R.
2024-01-01
Abstract
We propose a fast method to characterize inductive position sensors with zero physical prototypes. The technique is based on an in-house developed electromagnetic simulation tool which shows three orders of magnitude improvement with respect to the most widely used commercial software. This simulation software is used for producing synthetic data for training a machine learning surrogate model based on a neural network. In this way, the characterization of a sensor takes just milliseconds. This opens the possibility of devising a Design Support System (DesSS), a software that guide the user in sensor design for the best possible outcome, thereby saving development time.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
