磁控形状记忆合金动态建模及仿真研究

Dynamic model of magnetically controlled shape memory alloy

磁控形状记忆合金是一种新型功能材料,准确实用的动态模型的建立必然会为其进一步的应用奠定基础。然而由于其变形机理的复杂性,难以从其物理特性出发建模。因此以实验数据为基础,用最小二乘支持向量机回归建立动态模型,可以把磁控形状记忆合金动态建模问题转换为一个非线性小样本函数回归估计问题,并与BP神经网络在估计精度和泛化能力两方面进行了性能对比分析。仿真结果表明,最小二乘支持向量机在精度和泛化功能方面做到了最好的折衷,是用于磁控形状记忆合金回归分析建立其动态模型的一种很有效的方法。

The Magnetically Controlled Shape Memory Alloy is a new functional material, and the construction of concise and useful dynamic model is bound to lay a foundation for further application. However, because of its complexity of deformation mechanism, it is difficult to modeling from the physical properties. Therefore based on experimental data, using least squares support vector machine regression set up the dynamic model, the magnetic shape memory alloy dynamic modeling problem is changed into a nonlinear regression function of small sample problem, and comparative analysis was made by BP neural network in both estimating precision and generalization ability of performance. The simulation results show that the least squares support vector machine in the accuracy and the generalization of functions to do the best compromise, which is a very effective method for magnetic controlled shape memory alloy for the regression analysis and dynamic model building.