Hysteresis model of piezoceramics based on chaotic neural networks

A novel G-S chaotic neural network is proposed to resolve the hysteresis model of piezoceramics. The network has three layers: input layer, hidden layer and output layer. The input layer comprises the delay link, which makes the historical input capable to affect the current response. The learning algorithm is a process of chaos optimization, which can make the network avoid the local minima problem and false saturation phenomenon. The network can reduce the modeling error for the piezoelectric actuator of a nanometer positioning system. Experimental results proved validity of the algorithm.