Modeling and experimental validation of self-sensing high-frequency displacement of multi-dimensional discrete magnetostrictive actuator

A dual spool fuel valve has been developed for active control of combustion oscillations in aeroengines. The inner spool, directly driven by a multi-dimensional discrete magnetostrictive actuator (MDMA), regulates high-frequency fuel flow to meet the demands of active control system. Precise control of high-frequency fuel flow is critical for effective active control, necessitating accurate monitoring of the inner spool position. Traditional contact-based measurements, which rely on displacement sensors, are limited by integration complexity and intricate structural requirements. In contrast, displacement self-sensing methods for the MDMA offer a novel solution for monitoring the inner spool position. Therefore, this study presented a displacement self-sensing strategy employing four induction coils and developed a theoretical model based on the energy transfer process. However, the theoretical model exhibited low computational efficiency, and the induced current lagged behind the displacement, making it unsuitable for real-time high-frequency displacement calculation. To address this limitation, a real-time displacement self-sensing surrogate model (RDSSM) combining a two-layer multilayer perceptron network with a discrete wavelet transform algorithm was proposed. Experiments showed that the theoretical model had an average modeling error of 5.48 %, while the RDSSM achieved an error of 3.63 %. Furthermore, the closed-loop control experiments revealed an average relative error of 3.76 % between feedback displacement derived from the RDSSM and direct measurements, validating that the RDSSM can effectively replace traditional displacement sensors, enabling accurate self-sensing, feedback, and control of the inner spool position.