基于遗传算法的奥氏体SMA神经网络本构模型

Austenitic SMA neural network constitutive model based on a genetic algorithm

为了研究加/卸载速率对奥氏体SMA(shape memory alloy)本构的影响,以弥补Brinson等经典SMA本构模型的不足,进行了奥氏体SMA丝的力学性能试验。根据材性试验结果和神经网络、遗传两种智能算法理论,提出了两种速率相关型SMA本构模型,一种是BP网络本构模型,另一种是遗传算法优化BP网络本构模型。并对这两种本构模型进行了Matlab仿真模拟。SMA材性试验结果表明,循环稳定后,加/卸载速率对SMA的超弹性性能影响较大。Matlab仿真结果表明,以试验数据作为神经网络训练数据的遗传优化BP网络本构模型,其预测曲线与试验的应力-应变曲线吻合很好,预测误差很小,相比未优化BP网络本构模型具有更高的稳定性,是一种较为理想的速率相关型本构模型。

In order to study the impact of loading/unloading rates on the shape memory alloy （ SMA ） constitutive model, and to compensate for defects in classic SMA constitutive models, such as the Brison model, in this study, we performed austenitic SMA wire tests on its mechanical properties. Based on the material test results and two in-telligent algorithm theories, including neural network and genetic algorithms, we propose two types of rate-related constitutive models. One is the back-propagation （ BP ） network constitutive model, and the other is a genetic algo-rithm （ GA）-optimized BP network constitutive model. We simulated both constitutive models using Matlab software. The material test results show that after cycling stability, the loading/unloading rate has a significant impact on the super-elastic property of SMA wires. Using experimental data as training data in the BP network, the simulation re-sults show that the GA-optimized BP network model, whose prediction curve fits the experimental stress-strain curve very well and for which the calculation error is very small, has higher stability than the BP network constitutive model without optimization, and is an ideal rate-related constitutive model.