Aggogeri Francesco, Pellegrini Nicola
Department of Mechanical and Industrial Engineering, University of Brescia, Via Branze 38, 25123 Brescia, Italy.
Sensors (Basel). 2025 Aug 31;25(17):5371. doi: 10.3390/s25175371.
Piezoelectric actuators, widely used in micro-positioning and active control systems, show important hysteresis characteristics. In particular, the hysteresis contribution is a complex phenomenon that is difficult to model when the input amplitude and frequency are time-dependent. Existing dynamic physical models poorly describe the hysteresis influence of industrial mechatronic devices. This paper proposes a novel hybrid data-driven model based on the Bouc-Wen and backlash hysteresis formulations to appraise and compensate for the nonlinear effects. Firstly, the performance of the piezoelectric actuator was simulated and then tested in a complete representative domain, and then using the committee machine approach. Experimental campaigns were conducted to develop an algorithm that incorporated Bouc-Wen and backlash hysteresis parameters derived via genetic algorithm (GA) and particle swarm optimization (PSO) approaches for identification. These parameters were combined in a committee machine using a set of frequency clusters. The results obtained demonstrated an error reduction of 23.54% for the committee machine approach compared with the complete approach. The root mean square error (RMSE) was 0.42 µm, and the maximum absolute error (MAE) appraisal was close to 0.86 µm in the 150-250 Hz domain via the Bouc-Wen sub-model tuned with the genetic algorithm (GA).
压电致动器广泛应用于微定位和主动控制系统中，具有重要的滞后特性。特别是，当输入幅度和频率随时间变化时，滞后贡献是一种复杂的现象，难以建模。现有的动态物理模型难以描述工业机电一体化设备的滞后影响。本文提出了一种基于Bouc-Wen和间隙滞后公式的新型混合数据驱动模型，以评估和补偿非线性效应。首先，对压电致动器的性能进行了模拟，然后在一个完整的代表性域中进行了测试，随后采用了委员会机器方法。开展了实验活动，以开发一种算法，该算法结合了通过遗传算法（GA）和粒子群优化（PSO）方法导出的Bouc-Wen和间隙滞后参数进行识别。这些参数在委员会机器中使用一组频率簇进行组合。结果表明，与完整方法相比，委员会机器方法的误差降低了23.54%。通过遗传算法（GA）调整的Bouc-Wen子模型，在150 - 250Hz域内，均方根误差（RMSE）为0.42µm，最大绝对误差（MAE）评估接近0.86µm。
