Open AccessLong short term memory network is capable of capturing complex hysteretic dynamics in piezoelectric actuators
Author(s) -
Liu Yanfang,
Zhou Rui,
Huo Mingying
Publication year - 2019
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
ISSN - 1350-911X
DOI - 10.1049/el.2018.7490
Subject(s) - hysteresis , computer science , actuator , complex dynamics , topology (electrical circuits) , dynamics (music) , term (time) , network dynamics , complex network , network topology , range (aeronautics) , control theory (sociology) , artificial intelligence , engineering , physics , acoustics , control (management) , mathematics , electrical engineering , mathematical analysis , discrete mathematics , quantum mechanics , aerospace engineering , world wide web , operating system
This Letter demonstrates the capability of long short term memory (LSTM) network in capturing the complex hysteretic dynamics in piezoelectric actuators (PEAs). A LSTM network is constructed to model the PEAs' complex dynamics, which includes static hysteresis, creep, high‐order dynamics. The network is trained and evaluated by data sets of input–output pairs with different frequencies and amplitudes. Preliminary results show that, even for the simplest topology, namely one layer with one cell, the LSTM network provides a satisfactory precision in a wide frequency range. Thus, LSTM networks may provide a new approach to approximate the dynamics in complex engineering systems.