Heuristic modeling and inverse compensation of hysteresis in piezoelectric actuators based on time series similarity

Abstract

Hysteresis behavior of piezoelectric actuators degrades the accuracy and performance of the micro or nano positioning stage. As a remedy to this problem, a hysteresis model and a compensator based on time series similarity are proposed in this article. To generate this dynamic hysteresis model, geometric and time scale similarity of output displacement time series of piezoelectric actuators are studied. Furthermore, to compensate for hysteresis behavior, an inverse model based on time series similarity and an inverse model-based feedforward compensator are proposed. This novel modelbased compensator is validated against the Prandtl-Ishlinskii model-based one and is assessed for effectiveness and practicality.

A Hybrid Feedforward-Feedback Hysteresis Compensator in Piezoelectric Actuators Based on Least-Squares Support Vector Machine

“…Since the characteristic of the piezoelectric hysteresis depends not only on the amplitude but also on the frequency of input voltage signals, traditional rate-independent hysteresis models could yield errors subject to dynamic inputs with different frequencies [6]- [12]. To characterize the rate-dependent hysteresis, some models were put forward, such as the improved Preisach model [13], [14], improved Prandtl-Ishlinskii model [15], [16], and time series similarity model [17]- [19]. However, these models have a lot of parameters to be determined, which complicate the modeling process.…”

An adaptive weighted least square support vector regression for hysteresis in piezoelectric actuators

“…Hence, modeling the hysteresis using the traditional rate-independent hysteresis models such as the Preisach model [4], [5], Prandtl-Ishlinskii (PI) model [6], 7] and Krasnoselskii-Pokrovskii (KP) model [8] could yield errors subject to dynamic inputs with different frequencies. To characterize the rate-dependent hysteresis, some models were put forward, such as the improved Preisach model [9], [10], improved PI model [11], [12], and time series similarity model [13] - [15]. However, these models have a lot of parameters to be determined, which complicate the modeling process.…”