Research On Control And Hybrid Modeling Of Complex Hysteresis For Voice Coil Motor Under High Frequency Response

Voice coil motor (VCM) is a directly drive linear motor, which has the characteristicsof simple structure, small size and good force characteristics,and it avoids large inertia andlong mechanical hysteresis time in the traditional screw. VCM has the advantages of highfrequency response, high acceleration and high positioning accuracy, so it is widely used inthe field of ultra-precision machining, and semiconductor bonding.With the continuousimprovement of the technology requirements in these areas, voice coil motor is required towork in the state of high-frequency response more and more frequently.For example thebonding rate of the first generation bonder can achieve10lines/s， the bonding rate of thesecond-generation bonder can achieve25lines/s， and in the future, the bonding rate ofthe third-generation bonder will achieve40lines/s.However, a number of studies of domestic and foreign scholars show that there isobviously complex hysteresis phenomenon in the high frequency response in voice coilmotor. This complex hysteresis phenomenon seriously affects the positioning accuracy andthe tracking accuracy of voice coil motor, and it may produce chattering which causes thesystem instability. At this background, with the support of the National Natural ScienceFoundation, this paper have done deeply research of the complex hysteresis of thehigh-frequency response voice coil motor, and the main contents are as follows:1) The structure of the voice coil motor, and the principle of mechanical andelectromagnetic are introduced.The simple transfer function between output displacementand input voltage in voice coil motor is derived (mathematical model).This papersummarizes the current research status of the voice coil motor and points out thecomplexity hysteresis of voice coil motor in high frequency response.This phenomenondemonstrates a non-monotone form of hysteresis characteristics.2) This complex hysteresis is analysised in this paper. According to the non-monotonecharacteristics of SDH hysteresis model, combined with the least squares support vectormachine, a hysteresis hybrid modeling method which based on the least squares supportvector machine of SDH hysteretic model is put forward. The experimental data shows thatthis model can effectively describe the complex hysteresis of voice coil motor inhigh-frequency response, and it meets the requirements of complex hysteresis modeling;3) Because the hysteresis model expression of SDH is complex and difficult inpractical control applications, a neural network hysteresis hybrid modeling method basedon the constructing similar hysteresis is proposed. The so-called constructing similar hysteresis gives up the existing hysteresis model and builds a simple hysteresis modelwhich possesses the characteristic of non-monotonic. The experimental data shows that themodel accuracy of the neural network hysteresis hybrid model is further improve, and theexpression is simple and easy to achieve in the actual control;4) Regard the neural network hysteresis hybrid modeling method based on theconstructing similar hysteresis as the basis, a complex inverse model of neural networkbased on the constructing similar hysteresis is built. By combined with PID control, acomplex hysteresis inverse model controller of neural network is designed, in which thecontroller provides compensation and control for complex hysteresis of voice coil motor inhigh frequency response. At last, the control algorithm is verified at the control platform ofcSPACE voice coil motor, the experimental data shows that the control method caneffectively compensate for the complex hysteresis of voice coil motor, and it greatlyimproves the tracking accuracy of the voice coil motor and meets the requirements ofpractical application.The effectiveness of the control method is proved.