A neural network modeling and sliding mode control of self-sensing ionic polymer–metal composite actuator

This work reports on the development of integrated sensing and feedback control system for self-sensing ionic polymer–metal composite actuator. Integrated sensing is accomplished by crafting discrete sensing and actuation sections over a single device by patterning the surface electrodes. A control scheme and estimation technique is implemented for self-sensing feedback control that uses the electrode resistance change during deformation. Experiments are conducted to investigate the relation between the changes in electrode resistance of patterned sensor part to that of actual tip displacement of the device. Due to the large hysteresis and associated nonlinearity, artificial neural network is used as a computational tool to model their relation and to estimate the actual tip displacement of the device. The need for stable control to overcome nonlinearity and inherent back relaxation behavior of the material is accomplished using robust sliding mode controller. The experimental results based on the proposed method achieves good performance in terms of tracking control without the need for separate position sensor and makes the device to perform as a self-sensing actuator.