Magnetic shape memory alloy (MSMA) based actuators are extensively applied in the fields of precision manufacturing and micro/nano technology. Nevertheless, the inherent hysteresis in the MSMA-based actuator severely hinders its further application. In this study, the characteristics of hysteresis behavior under different input signals are investigated. Then, a nonlinear auto-regressive moving average with exogenous inputs (NARMAX) model based on a diagonal recurrent neural network (DRNN) is used to construct the rate-dependent hysteresis model. To improve the capability of characterizing the multi-valued mapping of the hysteresis loop, the play operator is adopted as the exogenous variable function of the NARMAX model. To verify the effectiveness of the proposed model, a series of comparisons are implemented. The experimental results show that the proposed NARMAX model based on the DRNN exhibits excellent modeling performance.

This study addresses the finite-time adaptive output feedback quantized control problem for nonstrict-feedback nonlinear systems subject to unknown hysteresis and time delays. An estimated inverse compensator (EIC) is constructed to mitigate hysteresis by resorting to a generalized Duhem model. A Lyapunov–Krasovskii functional is employed to deal with the uncertainties of time delays. However, the system performance deteriorates when the quantized signal is directly applied to the hysteretic system. To overcome this obstacle, this study proposes a new composite quantizer consisting of an adaptive state-estimation filter and a modified hysteretic quantizer, in which the former facilitates the state approximation by incorporating the feedback information of the system and the latter regulates the communication rate. With the proposed finite-time prescribed adaptive quantized control with an estimated inverse compensator (EIC–FPAQC) scheme, all solutions of the closed-loop systems are semiglobal practical finite-time stable and the tracking error can be ensured in a predefined accuracy. Finally, experiments are conducted on a piezoelectric-driven micropositioning stage to demonstrate the effectiveness of the proposed method.

In this study, the data-driven sliding-mode iterative learning tracking control problem of a piezoelectric-actuated micro-positioning (PAMP) stage is investigated. To improve the convergence performance of the data-driven sliding mode iterative learning control (DDSILC) method, a novel iteration-dependent parameter learning mechanism is proposed. Subsequently, an enhanced DDSILC (E-DDSILC) scheme is constructed. The novel parameter-learning mechanism is designed such that the tracking error in time-varying systems can converge to zero in the time domain at the final iteration, and to significantly improve the transient performance of the system. Additionally, the effect of control parameters on the convergence performance is analyzed, which enables the parameters to be adjusted reasonably and efficiently. Several comparison experiments are conducted on the PAMP stage to verify the effectiveness of the proposed control approach Note to Practitioners —Owing to the gradual industrial development toward the high-end manufacturing, many products, such as vascular robots and precision chips, have reached the micro/nano level of accuracy. The piezoelectric-actuated micro-positioning (PAMP) stage has been widely used in high-precision fields, such as fluorescence microscopy, nanoimprint lithography, and laser interferometry, owing to its fast response and ability to generate micro-nano displacements. However, because of the hysteresis and other nonlinear characteristics existed of the PAMP stage, advanced control algorithms are required to address the nonlinearity and satisfy the requirements of high-precision control. When the PAMP stage is used for precision manipulation tasks, such as nanolithography and micro/nanoimaging, the advanced control algorithms present the following restrictions: high dependence on offline models and the necessity to select parameters via trial-and-error method. These limitations render it difficult to implement control approaches. Hence, this study proposes an E-DDSILC scheme, which adopts the dynamic linearization to obtain the nonlinear information of the system. Unlike the conventional DDSILC method, the proposed scheme with enhanced iterative learning mechanism guarantees error convergence in the time domain. Furthermore, the effects of the main parameters on the transient and steady-state performances are investigated. As an offline model is not required for the proposed method and the effects of the parameters are explicit, the operation time is reduced and the feasibility of the controller in practical implementation is guaranteed.

Piezoelectric actuators (PEA) have been widely used in the ultra-precision manufacturing fields. However, the hysteresis nonlinearity between the input voltage and the output displacement, which possesses the properties of rate dependency and multivalued mapping, seriously impedes the positioning accuracy of the PEA. This paper investigates a control methodology without the hysteresis model for PEA actuated nanopositioning systems, in which the inherent drawback generated by the hysteresis nonlinearity aggregates the control accuracy of the PEA. To address this problem, a neural network self-tuning control approach is proposed to realize the high accuracy tracking with respect to the system uncertainties and hysteresis nonlinearity of the PEA. First, the PEA is described as a nonlinear equation with two variables, which are unknown. Then, using the capabilities of super approximation and adaptive parameter adjustment, the neural network identifiers are used to approximate the two unknown variables automatically updated without any off-line identification, respectively. To verify the validity and effectiveness of the proposed control methodology, a series of experiments is executed on a commercial PEA product. The experimental results illustrate that the established neural network self-tuning control method is efficient in damping the hysteresis nonlinearity and enhancing the trajectory tracking property.

Magnetic shape memory alloy (MSMA) based actuator plays an important role in the field of micro and nano-fabrication due to its large stroke. However, the inherent hysteresis of MSMA seriously affects the application prospects of MSMA-based actuator in the field of precision positioning. In this study, a composite model is proposed for hysteresis in the MSMA-based actuator by coupling an improved fractional-order Bouc-Wen model to a nonlinear auto-regressive moving average with exogenous inputs (NARMAX) model. Based on the proposed model, a neural network adaptive control method is then used to control the MSMA-based actuator, and the controller design takes into account the effect of time delay on the system performance to improve the positioning accuracy of the MSMA-based actuator. First, we fuse the composite NARMAX model with the canonical form of the nonlinear system to describe the MSMA-based actuator. Then, a control strategy is developed to address the impact of the unknown time delay for controller design and achieve the satisfactory positioning accuracy of the MSMA-based actuator. By using Lyapunov theory, the tracking error is proven to be asymptotic convergence. Experimental studies are provided to validate the effectiveness of the proposed modeling and control schemes.