RD-CNN for driving a 2-D conveyor belt via memory shape alloys

摘要

The paper proposes a two-dimensional conveyor belt controlled by using an analog neural processing approach. A particular Cellular Neural Networks (CNN), named Reaction―Diffusion CNN (RD-CNN), is adopted to generate wave propagation phenomena. These waves can propagate on the conveyor belt plane (an elastic membrane) moving an object on it. The adopted actuator is a shape memory alloy such as nitinol. A neural network identification approach was adopted to characterize the membrane model. It is shown how both thermal and timing nitinol behaviors are very similar to the slow―fast dynamics exhibited by a RD-CNN allowing an appropriate driver system to be designed.