Underlying memory-dominant nature of hysteresis in piezoelectric materials

Abstract

Although the existence of nonlocal memories in hysteresis behavior of piezoelectric materials has been demonstrated, their detailed and thorough properties have yet to be revealed. Along this line, we disclose and demonstrate the underlying memory-dominant nature of hysteresis, and characterize its important properties that must be considered for the accurate prediction of hysteresis trajectory in piezoelectric materials. More specifically, the concept of recording the turning points, targeting the previously recorded turning points, curve alignment, and wiping-out effects at these points are introduced as the basic intellectual properties of hysteresis nonlinearity. A constitutive memory-based mathematical modeling framework is then developed and trained for the precise prediction of a hysteresis path for arbitrarily assigned input profiles. Utilizing a piezoelectric-driven actuator, it is experimentally demonstrated that if the number of memory units is sufficiently selected, model response in the prediction of a hysteresis track is significantly improved.