SMARC变形控制及神经网络建模

在复合材料层板中偏离中性层的位置铺设镍钛形状记忆合金丝,然后通电加热镍钛丝,使其在形状回复过程中产生巨大的回复力,从而使层板结构发生弯曲变形.由于形状记忆合金的非线性以及合成的智能材料结构的应力、应变分布和形状变化规律都非常复杂,尝试引入人工神经网络来建立以可控参量(电流强度)为输入变量、易测参量(稳态挠度)为输出变量的模型,所建模型的预测数据与实验数据之间符合得较好,相对误差小于9%.

Study of active shape control of SMA reinforced composites and the neural network model

NiTi SMA wires were embedded in the composite laminate beyond the neutral layer, and by electric current, they are heated up to produce the large recovery force. Consequently, the bending deformation of SMA reinforced structure is achieved. Due to the nonlinear behavior of SMA and the complexity of the principles of stress, strain and shape of the composite structure embedded with SMA wires, use the artificial neural network to establish the model of steady-state deflection-intensity of electric current. Eventually, the relative error between the expected date of the model and the experimental data is less than 9%.