MFC致动器的动态迟滞模型辨识及补偿控制EI北大核心CSCD

压电宏纤维(marco fiber composite,MFC)具有柔性好、变形能力强的优点,但MFC致动器驱动的柔性臂的迟滞非线性严重影响系统定位精度。提出一种具有非对称性的改进Prandtl-Ishlinskii(PI)迟滞模型,解决经典PI迟滞模型的缺陷(对称性);该模型基于经典PI迟滞模型,叠加一系列不同权重、不同阈值的双边死区算子获得,基于最小二乘法的迟滞模型辨识结果表明,改进PI迟滞模型对MFC致动器的迟滞建模误差从PI迟滞模型误差的16.06%降到5.58%。另外,建立系统的离散传递函数模型来描述系统的线性动态特性,并与改进PI迟滞模型串联得到组合模型,解决纯迟滞模型仅能描述低频、准静态情况下的迟滞特性问题。在前馈补偿下,对MFC致动的柔性臂进行正弦波轨迹跟踪试验,测得补偿后实测位移与期望跟踪位移基本吻合,跟踪精度达到93.62%以上。试验结果证明,所提出的改进PI迟滞模型、离散传递函数模型及补偿方法的有效性。     

压电致动的谐振式水下柔性结构动态迟滞建模及前馈补偿

具有运动灵活且操作方便优点的宏压电纤维复合材料(macro fiber composite, MFC)致动水下柔性结构广泛应用于水下仿生推进和变形控制系统中，但是MFC的迟滞非线性严重影响了系统的定位精度和操控性能。提出了一种改进Prandtl-Ishlinskii(PI)静态迟滞和传递函数动态模型串联的复合式模型来描述MFC致动水下柔性结构谐振状态下的动态迟滞行为。首先基于所提出水下结构的准静态迟滞特性辨识得到改进PI迟滞模型参数，然后通过传递函数串联馈通环节的动态模型捕捉MFC致动柔性结构的水下谐振特性。实验结果表明所建立的复合式动态迟滞模型能够很好地描述MFC致动水下柔性结构在谐振状态下的动态迟滞行为，并且在固有频率附近一定带宽范围内仍具有较高准确性。基于复合式逆模型的前馈补偿下，水下柔性结构在谐振状态下跟踪正弦轨迹的实测位移与期望位移基本重合，补偿后二者线性度较高，显著提升了MFC致动柔性结构谐振状态下的动态定位和跟踪精度，证实了所提出动态迟滞模型和补偿方法的有效性。     

压电驱动器的非对称迟滞模型

为了补偿压电驱动器的非对称迟滞,提出了改进型Maxwell迟滞模型.Goldfarb提出的经典Maxwell迟滞模型由多个基础单元并联叠加而成,其基础单元为单个弹簧-物块单元,迟滞特性为平行四边形,故只能描述对称迟滞.为了能描述压电驱动器非对称迟滞,提出了基础单元迟滞特性为梯形的模型.为了简化算法程序,将梯形单元优化为两个三角形单元组合而成.为了验证该模型,以压电工作台为实验对象,运用该迟滞模型的逆模型进行迟滞补偿控制.单独的迟滞逆模型前馈开环控制实验结果表明,位移跟踪相对误差从7.37％降到了1.56％,输入输出基本呈线性关系.逆模型结合PID复合反馈闭环控制实验结果表明,位移跟踪相对误差进一步降低到0.53％,输入输出呈很好的线性关系.这表明本文所建立的迟滞模型能很好地描述压电驱动器非对称迟滞特性.     

基于PI迟滞模型的压电驱动器自适应辨识与逆控制

压电陶瓷驱动器的迟滞非线性特性严重影响了其跟踪定位精度,甚至引起闭环系统失稳.本文采用经典PI模型描述压电驱动器的迟滞非线性,利用自适应投影算法对PI模型的权向量进行在线辨识,并与传统的最小二乘辨识方法进行比较.迟滞PI模型的优点是模型存在解析逆,因此本文对压电驱动器采用自适应逆跟踪控制,利用驱动器的输出位移与参考位移之差使用自适应投影算法在线辨识PI模型的权向量,并计算PI逆模型的权向量和阈值,最终得到要输入的电压值.最后实验结果表明自适应逆跟踪控制比传统的逆模型跟踪控制精度提高了49.8%.     

基于Prandtl-Ishlinskii模型的智能悬臂梁迟滞补偿与自适应复合控制研究

随着智能结构在当今工程的应用越来越广泛,智能结构的迟滞特性受到越来越多的关注。这种迟滞特性不但影响系统的输出精度,甚至威胁系统的稳定性。为了解决这一问题,本文提出一种自适应复合控制策略:首先运用最小二乘算法在线实时辨识智能结构PI(Prandtl-lshlinskii)正模型,模拟智能结构的迟滞特性;基于智能结构PI(Prandtl-lshlinskii)正模型建立智能结构的PI逆模型,实现智能悬臂梁迟滞特性的补偿;同时,将智能结构的输出作为反馈信号,采用最小方差自校正控制律在线设计反馈增益,达到构成智能结构闭环控制系统的目的。通过数值仿真,对比智能结构在有无迟滞补偿两种情况下的振动控制效果,探索智能结构迟滞特性对智能结构的振动控制效果的影响规律,并通过实验验证两种情况下智能结构振动控制结果,得出迟滞特性影响智能结构振动控制效果的有效结论。结果表明:在有迟滞补偿情况下智能结构的振动控制效果优于无迟滞补偿情况下的控制效果。     

基于三段PI模型的压电驱动器迟滞补偿方法

迟滞非线性降低了压电陶瓷驱动器对期望位移的跟踪精度,为解决该问题,本文通过二次规划寻优算法,基于最小均方误差准则,以压电陶瓷驱动器的运动速度规律为依据,对压电陶瓷运动速度规律不同的迟滞段分别进行PI建模.建模结果表明,相较于传统PI模型,本文提出的三段PI模型能精确地描述压电陶瓷迟滞曲线的非奇对称性.在对压电陶瓷三段PI建模的基础上,计算出压电陶瓷三段PI逆模型的阈值向量与权系数向量.通过建立的三段PI逆模型对压电陶瓷进行迟滞补偿控制,实验结果表明,与传统PI逆模型迟滞补偿控制相比,三段PI逆模型迟滞补偿控制方法将压电陶瓷对期望位移的跟踪精度提高了81.3%.     

聚吡咯驱动器迟滞非线性特性探究

在仿生学领域中,聚吡咯驱动器因具有结构简单,适应性强,抗电磁干扰等优点得到广泛应用;通过对驱动器的传递函数进行建模分析,得出其存在迟滞非线性特性,且该特性会造成驱动器跟踪定位精度低,可控性差,所以采用PI模型对驱动器进行迟滞建模,并研究其迟滞特性.首先,对传递函数模型中得到的输出位移进行误差分析,得知驱动器存在迟滞非线性特性,将系统模型分为纯迟滞模型及传递函数模型两部分;其次,利用PI模型的算法对不同输入频率信号下的系统分别构建迟滞模型,并通过分析基于迟滞模型的系统模型的输出结果,得出驱动器存在迟滞非线性特性,且该特性与输入信号频率无关;最后通过比较试验测量输出数据与仿真结果,验证结论的有效性.     

计及动态负载的气动肌肉驱动器迟滞模型

气动人工肌肉是一种新型的柔顺性驱动器,但由于采用硅胶等非线性材料而存在明显的迟滞现象,并且其滞环参数随着负载的不同而变化。本文基于广义Prandtl-Ishlinskii迟滞模型,引入气动肌肉的负载参数,建立了计及动态负载的迟滞模型Load-dependent generalized Prandtl-Ishlinskii(LG PI)。在0～100 N的负载范围内进行实验,验证了该模型能较好地描述气动肌肉在动态负载作用下的不对称迟滞特性,其滞环回路位置误差最大不超过1 mm。将LGPI模型的逆作为PID控制的前馈环节,以控制气动肌肉驱动的机械单指灵巧手手指,在不同负载下进行角度跟踪实验。与无前馈的PID控制实验结果对比分析表明,增加基于LGPI模型的前馈环节能提高控制系统的响应速度和控制精度。     

压电作动器的迟滞非线性建模和实时补偿控制仿真

压电作动器的迟滞非线性不同程度的影响了作动器的性能,降低了系统的稳定性,甚至使得系统不稳定,尤其在振动主动控制应用中,要求系统具有极强的实时性,否则由于相位滞后控制效果将会受到严重影响.在研究了各种迟滞非线性模型和补偿算法的基础上,采用PI迟滞算子对压电作动器建立一个纯现象的模型来准确描述其迟滞现象,参数的线性不等式约束保证了求解的唯一性和模型的可逆性.在此基础上利用PI逆模型设计补偿器来补偿迟滞非线性.仿真结果证明了控制算法的正确性和有效性,并有效抑制了迟滞的影响,保证了理想的跟踪精度.     

压电式二维微定位平台的率相关迟滞建模

为了描述压电式多维微定位平台的率相关迟滞非线性特性,提出了一种基于Hammerstein模型的建模方法。以一种二维微定位平台为对象,平台动态模型是由静态迟滞非线性部分和一个线性动态系统部分串联组成。静态非线性部分由改进的Prandtl-Ishlinskii模型描述,线性动态系统部分由外因输入自回归模型(ARX)模型描述,并给出了模型参数辨识方法。为了验证所建立的Hammerstein模型有效性,搭建了实验平台进行实验验证。实验结果表明,对平台施加不同频率电压信号,由Hammerstein模型得到的预测位移和实测位移相对误差范围为1%~5%,预测位移与实测位移接近,说明所建立的模型能精确描述微定位平台的率相关迟滞特性。     

A hysteresis model for a stacked-type piezoelectric actuator

The hysteresis nonlinearity in piezoelectric materials brings difficulties in controlling the systems. In order to mitigate the effect of hysteresis, such nonlinearity needs to be characterized and modeled under different load circumstances. For this purpose, the actuator is modeled in terms of a mass-spring-damper system utilizing the stop operator as one of the operators of the Prandtl-Ishlinskii (PI) model. Merging the structural model with the nonlinear hysteresis model, we observe that the results demonstrate better correspondence to the measured output compared to that of the classical PI model for a wide range of working conditions, i.e., different input frequency and amplitude.     

一种改进的形状记忆合金驱动器自适应控制算法的初步研究

针对形状记忆合金驱动器的非线性迟滞特性,提出了一种改进的自适应控制算法。该算法以MKP逆模型为前馈补偿器,通过选择性自适应律调整模型参数,从而实现动态磁滞补偿,以达到实时准确控制。采用仿真实验对其性能进行了初步验证。结果表明,该算法具有收敛速度快、运算负荷小、抗噪能力强等优点,适合于实时控制应用。     

Hysteresis in a carbon nanotube based electroactive polymer microfiber actuator: numerical modeling

Hysteretic behavior is an important consideration for smart electroactive polymer actuators in a wide variety of nano/micro-scale applications. We prepared an electroactive polymer actuator in the form of a microfiber, based on single-wall carbon nanotubes and polyaniline, and investigated the hysteretic characteristics of the actuator under electrical potential switching in a basic electrolyte solution. For actuation experiments, we measured the variation of the length of the carbon-nanotube-based electroactive polymer actuator, using an Aurora Scientific Inc. 300B Series muscle lever arm system, while electrical potentials ranging from 0.2 V to 0.65 V were applied. Based on the classical Preisach hysteresis model, we presented and validated a numerical model that described the hysteretic behavior of the carbon-nanotube-based electroactive polymer actuator. Inverse hysteretic behavior was also simulated using the model to demonstrate its capability to predict an input from a desired output. This numerical model of hysteresis could be an effective approach to micro-scale control of carbon-nanotube-based electroactive polymer actuators in potential applications.     

High-bandwidth control of a piezoelectric nanopositioning stage in the presence of plant uncertainties

Inversion-based feedforward techniques have been known to deliver accurate tracking performance in the absence of plant parameter uncertainties. Piezoelectric stack actuated nanopositioning platforms are prone to variations in their system parameters such as resonance frequencies, due to changes in operating conditions like ambient temperature, humidity and loading. They also suffer from nonlinear effects of hysteresis, an inherent property of a piezoelectric actuator; charge actuation is applied to reduce the effects of hysteresis. In this work, we propose and test a technique that integrates a suitable feedback controller to reduce the effects of parameter uncertainties with the inversion-based feedforward technique. It is shown experimentally that the combination of damping, feedforward and charge actuation increases the tracking bandwidth of the platform from 310 to 1320?Hz.     

A new simple asymmetric hysteresis operator and its application to inverse control of piezoelectric actuators

Piezoelectric actuators (PEAs) are commonly used as micropositioning devices due to their high resolution, high stiffness, and fast frequency response. Because piezoceramic materials are ferroelectric, they fundamentally exhibit hysteresis behavior in their response to an applied electric field. The positioning precision can be significantly reduced due to nonlinear hysteresis effects when PEAs are used in relatively long range applications. This paper describes a new, precise, and simple asymmetric hysteresis operator dedicated to PEAs. The complex hysteretic transfer characteristic has been considered in a purely phenomenological way, without taking into account the underlying physics. This operator is based on two curves. The first curve corresponds to the main ascending branch and is modeled by the function f1. The second curve corresponds to the main reversal branch and is modeled by the function g2. The functions f(1) and g(2) are two very simple hyperbola functions with only three parameters. Particular ascending and reversal branches are deduced from appropriate translations of f(1) and g(2). The efficiency and precision of the proposed approach is demonstrated, in practice, by a real-time inverse feed-forward controller for piezoelectric actuators. Advantages and drawbacks of the proposed approach compared with classical hysteresis operators are discussed.     

基于光控压电混合驱动悬臂梁独立模态控制

本文提出利用镧改性锆钛酸铅（PLZT）的光电效应,将PLZT作为电动势源来驱动压电作动器,从而实现光控板壳结构的振动控制。基于光控压电等效电学模型建立了光控压电混合驱动的数学模型,并进行了实验验证。为了实现光控悬臂梁的独立模态控制,针对悬臂梁结构,设计了正交模态传感器/作动器表面电极形状函数。提出PLZT与压电作动器正/反接控制的激励策略,并结合速度反馈定光强控制的控制算法,利用Newmark-β法对不同光照强度下悬臂梁的动态响应进行了数值仿真分析。分析结果证明了本文所设计的模态传感器/作动器及针对光控压电混合驱动提出的控制策略的正确性。     

Numerical and experimental characterizations of longitudinally polarized piezoelectric d 15 shear macro-fiber composites

This contribution presents original numerical and experimental characterizations for prototyped longitudinally polarized piezoelectric d ₁₅ shear macro-fiber composites (MFC). The numerical characterization consists of a finite element (FE) simulation based on a representative volume element. It implements an enthalpy-based homogenization method (EBHM), recently proposed by the authors, as an extension of the so-called strain energy method to orthotropic piezoelectric fiber-reinforced composites. The numerical validation is carried out on a previously assumed layout of shear MFC. Later on, the EBHM is used to get the effective electromechanical material parameters of the shear MFC actual layout. These parameters are further validated experimentally through their use in the FE simulation of an original actuation benchmark that is proposed for the manufactured shear MFC experimental characterization. The latter is based on low-frequency (quasi-static) displacement measurements where the shear MFC serves as a voltage-driven actuator. Due to the small overall dislocation, a laser vibrometer is used for the measurements. The comparison of experimental and numerical results shows a reasonably good agreement and a nonlinear actuation response is observed. This work’s major outcomes are the experimental validation of the EBHM and the actuation functional operability of the manufactured longitudinally polarized piezoelectric d ₁₅ shear MFC. This opens the possibility for their application as actuator and sensor of shear-induced bending and torsion for vibration, shape and health control, or as a transducer for energy harvesting.     

Nonlocal hysteresis function identification and compensation with neural networks

This paper discusses the on-line identification of nonlocal static hysteresis functions, which are encountered in mechanical friction, magnetic materials, and piezoelectric actuators and cause problems in the design of controllers. In this article a new compensation method for friction in presliding regime is introduced that is based on the simplified Leuven Friction Model and on technology borrowed from neural networks. We present a solution how to identify the hysteresis caused by the friction and how to use this identified model for the compensation of the friction effects. The solution can be used for on-line identification and compensation. Results from both simulations and experiments will be shown.