Research On High-precision Actuator Of Magnetically Controlled Shape Memory Alloy

Magnetically Controlled Shape Memory Alloys（MSMA）are known as potentialmaterial for a new generation of actuators. As actuators, it meet the requirements oflarge stroke and high frequency in modern industrial. However, a strong hysteresisbetween the input and output of MSMA actuators make it different in application ofhigh precision and ultra-high-precision positioning field. In order to improve controlaccuracy of MSMA actuators, supported by Wuhan city Science and TechnologyResearch Project (201010621231), we carry out related research work. The mainresearch contents is as follows:In order to study MSMA materials deformation mechanism, the multi fieldcoupling device is used to measure the properties of NiMnGa. Based on the measureddata, the relationship between magnetic field, the external force and temperature withMSMA deform is analyzed. By loading the dynamic magnetic field, the nonlinearhysteresis curve of the material is got.After the researches on basic properties of MSMA, a magnetic circuit is designedto generate magnetic field through theoretical calculations, the feasibility of themagnetic circuit is verified by ANSYS software. On this basis, the overall structure ofMSMA actuator is designed.The traditional modeling methods have relatively large error, in order to improvethe modeling accuracy, artificial neural network model is adopted. To solve the problemthat neural network can’t approach one to many mapping, double sigmoid function isused as the activation function in the hidden layer of the neural network model,Levenberg-Marquardt algorithm is adopted as the learning algorithm to improve thetraining speed. By writing cycle program, best hidden layer neuron number is got. Afterthat，artificial neural network model formula is established by weights and thresholds.MSMAs have a strong hysteresis between input and output, which make MSMA actuators a low control accuracy. In order to improve the control accuracy, PID theory,inverse compensation theory and fuzzy control theory are studied. Five kinds of controlways is used to reduce the effect of hysteresis nonlinear. Results based on simulinkshow that the classical PID control method and the inverse feedforward compensationcontrol method have relatively large error. The hybrid control method of chassical PIDand inverse feedforward compensation can greatly improve the control accuracy. FuzzyPID control method improve control accuracy for further on the basis of the hybridcontrol method. Compared with the fuzzy PID control method, The hybrid controlmethod of fuzzy method and inverse model with a0.1163um maximum error hassmaller overall error, so this control method is the best way to improve MSMA actuatorcontrol accuracy.