Application Of BP Neural Network Control Algorithm In AFM

Atomic force microscope(AFM)has been playing an important role and one of the principal research tools in the field of nanometer.At present,the commercial AFM control system uses the proportional integral(PI)control algorithm to control the piezoelectric ceramic actuator precisely.Piezoelectric ceramics have complex nonlinear characteristics,is difficult to precise control with simple PID controller,which not only affects the scanning speed of AFM,but also affects its testing accuracy.Thereby,many teams optimize AFM systems by improving piezoelectric ceramic actuators or control algorithms.In comparison,optimizing the controller algorithm is low cost and wide applicability.This paper analyses the current development of AFM,PID intelligent control algorithm based on BP neural network optimization to majorization AFM control system were studied.The intelligent control algorithm show the advantages of parameter self-learning,self-tuning and self-adaptation after introduction of BP neural network.AFM can acquire the self-learning ability,enhancing the self-adaptability when applying the intelligent control algorithm.Precision control of the piezoelectric ceramics can improve the quality of the scanned image,make the system more robust,and enhance the anti-interference ability of the AFM system.The simulation system of AFM can provide convenient and fast first-hand information for its research and application.The research on this aspect is also a focused research direction in micro-nano field.The simulation system of AFM can provide convenient and fast first-hand information for AFM research and application,the research on this aspect is also a focused research direction in micro-nano field.This study establish AFM Imitation products platform,firstly analyses the basic working principle and its system of actuators in nonlinear and time-varying,on this basis design modeling and use Simulink combine with S function established AFM in "contact" and "tap" two kinds of working mode,input signal waveform simulation samples surface morphology,finally by scanning the input signal waveform verified simulation platform.