This paper studies the vibration control of cylindrical shell panel. Shape memory alloy (SMA) wires are adopted as actuators and neural network training method is applied to enable SMA to output desired force. The proposed SMA actuators are fixed on the cylindrical shell panel and generate contracting forces while be heated electronically. The SMA actuators need to generate forces opposite to the external loading, and thus they can suppress the effect of external force. However, as SMAs present nonlinear relationship between temperature and force, it is difficult to output the desired stress profile. In this research, the hysteresis characteristics of SMAs are fitted by neural networks. The neural network model of SMA plant model is established based on the temperature input and force output; and the inverse model is generated with the force as input and temperature as output. The desired forces are obtained based on these neural network models. To validate the effectiveness of SMA actuators, the vibration response of cylindrical shell panel is analyzed with modal expansion method. The modal response under the controlling of SMA actuators is calculated based on the modal dynamic equation. The results show that SMA actuators controlled by the neural network method are effective to suppress the vibration of cylindrical panel shell. Primary experiments were performed to verify the proposed neutral network method. The results show that the SMA wire actuator generated desired force profile while heating using the neutral network method.

This content is only available via PDF.
You do not currently have access to this content.