The magnetic shape memory alloy (MSMA)-based actuator, as a new type of actuator, has a great application prospect in the micro-precision positioning field. However, the input-to-output hysteresis nonlinearity largely hinders its wide application. In this paper, a Takagi–Sugeno fuzzy neural network (TSFNN) model based on the modified bacteria foraging algorithm (MBFA) is innovatively utilized to describe the complex hysteresis nonlinearity of the MSMA-based actuator, and the parameters of TSFNN are optimized by the MBFA. The TSFNN is a combination of the fuzzy-logic system and neural network; thus, it has the capability of approximating the nonlinear mapping function and self-adjustment and is suitable for hysteresis modeling. The MBFA, which can obtain better optimization values, is employed for the parameter identification procedure. To demonstrate the effectiveness of the proposed model, a TSFNN based on the gradient descent algorithm (GDA) is used for comparison. Experimental results clearly show that the proposed modeling method can accurately describe the hysteresis nonlinearity of the MSMA-based actuator and has significance for its future application.

中文翻译：

---

基于改进细菌觅食算法的磁形状记忆合金驱动器Takagi-Sugeno模糊神经网络滞后建模

基于磁性形状记忆合金（MSMA）的执行器作为一种新型执行器，在微精度定位领域具有广阔的应用前景。但是，输入至输出磁滞非线性极大地阻碍了其广泛应用。本文创新地利用了基于改进的细菌觅食算法（MBFA）的Takagi-Sugeno模糊神经网络（TSFNN）模型来描述基于MSMA的执行器的复杂滞后非线性，并通过优化TSFNN的参数。 MBFA。TSFNN是模糊逻辑系统和神经网络的结合；因此，它具有逼近非线性映射函数和自我调整的能力，适用于滞后建模。可以获取更好的优化值的MBFA用于参数识别过程。为了证明所提出模型的有效性，将基于梯度下降算法（GDA）的TSFNN用于比较。实验结果清楚地表明，所提出的建模方法能够准确地描述基于MSMA的执行器的磁滞非线性，并对其未来的应用具有重要意义。