Back propagation neural networks based hysteresis modeling and compensation for a piezoelectric scanner

As the actuator of a common atomic force microscopy (AFM), a piezoelectric scanner has many advantages than other actuators, such as high precision of displacements on the nanoscale, high efficiency of electromechanical coupling, rapid response and so on. However, hysteresis nonlinearity of a piezoelectric scanner affects the positioning of the scanner and image quality of an AFM system. In this paper, a modeling method based on Back Propagation Neural Networks (BPNN) is proposed to compensate for hysteresis behavior. In particular, considering memory characteristics and frequency dependence of the hysteresis effect, firstly, we utilize a two hidden layers BPNN consisting of an input layer including the frequency and a section of the input voltage, two hidden layers, and an output layer to model for hysteresis. Subsequently, a method based on cubic spline interpolation is proposed to compensate for hysteresis behavior. Experiment results demonstrate the high precision of the obtained hysteresis model and the good performance of the proposed compensation method.