Investigation Of High-precision Displacement Control Methodology For Magnetically Controlled Shape Memory Alloy Actuators

As a new kind functional material, magnetic shape memory alloy has a short history ofdevelopment, but compared with other functional materials, it has obviously advantages as:larger output displacement, rapid frequency response, larger output stress and easy control.These advantages make people believed that the magnetic shape memory alloy has a brightfuture in the field of micro driving and micro positioning, and it has the opportunity to play arevolutionary role in this field. Due to the short development history of magnetic shapememory alloy actuator, the research on high accuracy control algorithm of magnetic shapememory alloy actuator has just begun. In this paper, the relationship between input andoutput of magnetic shape memory alloy actuator is studied and analyzed, the actuator’smodel and its inverse model are established and corresponding high accuracy positioncontrol algorithm are proposed. The research in this paper has laid the foundation for thefurther application of magnetic shape memory alloy actuator.In this paper, the deformation mechanism of magnetic shape memory alloy and theworking principle of magnetic shape memory alloy actuator are analyzed, hysteresis betweeninput magnetic induction density and output displacement of magnetic shape memory alloyactuator is studied, and then the hysteresis compensation is determined to be the maincontrol method in this paper. In order to establish the model of magnetic shape memory alloyactuator, first, neural network model is adopted. To solve the problem that neural networkcan’t approach one to many mapping, double Sigmoid function is used as the activationfunction in the fist hidden layer of the neural network model, momentum BP algorithm isadopted as the learning algorithm of the neural network. Second, in order to establish ahysteresis model which can describe more complex hysteresis, a KP hysteresis model isestablished in this paper. Improved gradient correction algorithm and variable step-sizerecursive least square estimation algorithm are used to identify the weighting parameters ofthe KP operators. The experiment results show that the proposed KP model has the ability todescribe the complex hysteresis loops.Based on the proposed KP model, the inverse hysteresis model of magnetic shapememory alloy actuator is established in this paper, then by using the proposed inversehysteresis model as the feed-forward controller, feed-forward control of magnetic shapememory alloy actuator is realized. The simulation results show that the proposedfeed-forward controller in this paper can compensate the hysteresis effectively and highlyimprove the control accuracy of the system. In order to improve the control accuracy, classical PID feed-back control is added to thefeed-forward controller then the hybrid control system of magnetic shape memory alloyactuator is composed, with the hybrid control, the magnetic shape memory alloy actuator canget high position control accuracy. At last, a self tuning PID controller based on RBFnetwork is used to replace the classical PID controller as the feed-back controller, itenhances the self-adaptive ability of the system, and it makes the control accuracy improvedagain. The simulation results show that the proposed hybrid control algolithm can realize thehigh-precision displacement control of magnetic shape memory alloy actuator.