基于杂草算法的超磁致伸缩作动器耦合模型识别

考虑磁滞损耗、动态应力等因素的超磁致伸缩作动器磁滞模型可有效揭示电-磁-机-热多场耦合效应，但准确识别其非线性模型往往存在较大困难。智能杂草算法具有激烈的竞争机制和较强的搜索能力，可用于解决作动器多目标物理参数辨识问题。传统算法的种子数量以线性方式产生且分布方差与适应度缺乏联系，极大地影响了算法收敛速度和模型识别精度。为此，提出一种非线性繁殖和分布的混合改进杂草算法，并将其应用于超磁致伸缩作动器模型识别。实验表明：改进算法具有较强的噪声抑制能力，能精确辨识含有噪声扰动的作动器磁滞非线性模型物理参数；模型预测值和实验数据误差较小，所识别参数可使磁滞非线性模型较为全面地描述作动器多场耦合机理和动态特性。
The hysteresis model of giant magnetostrictive actuator, considering the hysteresis loss and dynamic stress, can comprehensively reveal the electric, magnetic, mechanical and thermal multi-field coupling effect. It is often, however, difficult to accurately identify the nonlinear model by the experiment. The intelligent IWO (invasive weed optimization) with a fierce competition mechanism and strong search ability is very suitable for solving the problem of multi-objective physical parameters identification. Nevertheless, the number of seeds is linearly generated in traditional IWO and the distribution variance is lack of adaptability as well, which greatly affects the algorithm convergence speed and model recognition accuracy. Therefore, an improved algorithm with nonlinear propagation and distribution was proposed and applied to the model parameters identification of giant magnetostrictive actuator. The experiment exhibits that the improved algorithm has stronger noise suppression ability, which can accurately identify the physical parameters of the hysteresis nonlinear model with noise signal, and the errors between model predictions and experimental data are much smaller, thus the identified parameters can make the hysteresis nonlinear model comprehensively describe the actuator multi-field coupling mechanism and dynamic characteristics.