Research on SMA motor modelling and control algorithm for optical image stabilization

Abstract

Cameras play an increasingly important role in mobile devices and intelligent systems. With the development of technology, the resolution of the camera continues to improve and the images become much clearer. However, image quality is easily affected by lens shake. Optical stabilization technology improves the shooting stability as well as image quality of mobile devices by overcoming image blur which is caused by lens shake. In the current camera modules of smart devices, the voice coil motor (VCM) drives the lens movement. However, the traditional VCMs contain permanent magnets, thus generating stray magnetic fields and affecting the quality of captured images. Compared with VCM technology, shape memory alloy (SMA) motors driven by shape memory alloys have the advantages, such as lighter weight, higher accuracy, greater thrust, and lower energy consumption. But there are still some issues with the current SMA motors. To begin with, there is a control problem of SMA wire, and the current control algorithms are difficult to achieve high-precision control of SMA wire. What’s more, the strain range of SMA wires used in current SMA motors is limited, so their application in certain fields are restricted. A SMA motor with a larger shrinkage of SMA wire was designed, which could be used in larger lenses and whose application range was extended. Given the hysteresis effect of SMA, a Hammerstein-like model was established to describe the dynamic hysteresis characteristics of SMA. The Prandtl Ishlinskii model was used to describe the static hysteresis model of SMA, and the Nonlinear AutoRegressive network with eXogenous inputs (NARX) model was used to describe the dynamic model of SMA and predict its properties. A composite control scheme based on inverse compensation was designed to eliminate the dynamic hysteresis phenomenon of SMA, using the PI inverse model as the feed-forward controller and BP-FOPID as the feedback controller. In a gesture to address the problem of multiple fractional-order PID parameters and the difficulty of parameter tuning, the BP neural network and fractional-order PID controller were combined, and the BP neural network was used to rectify the five parameters of the fractional-order PID, which makes up the BP-FOPID. The performance and stability of the SMA motor control system were improved by integrating the feed-forward and feedback control strategy. And in the final experiment, the SMA motor experimental setup designed, the experimental platform built, the composite controller used to achieve the precise control of the SMA motor, the precise control of the SMA wire could be realized by controlling the duty cycle of the PWM wave, so that the SMA motor could effectively reduce the lens shake and improve the clarity and stability of the image. Through studying the structure and control method of the SMA motor, an SMA motor is designed and a corresponding control scheme is formulated, which makes the motor have a good anti-shaking effect, thus improving the image quality of the lens. Hopefully it can be applied to more fields.