Research On Methodology Of Structure Optimization,Modeling And Control Of Hysteresis In Giant Magnetostrictive Radial Micro-Feed Mechanism

Non-circular pin-hole in piston can obviously optimize stress distribution,improve lifetime and performance.However,the particularity of non-circular pin-hole in piston increases difficulties of precision machining.Therefore,we investigated an intelligence boring system based on the giant magnetostrictive actuator(GMA),to realize precision machining of piston non-circular pin-hole.In this study,the key technologies in GMA intelligence boring system were researched in detail,and the main innovation research works can be summarized as follow:optimization of GMA intelligence boring bar by whole model,rate-dependent hysteresis nonlinearity model for GMA intelligence boring system,inverse-model based iterative learning control(ILC)for GMA intelligence boring system and boring experiment of piston non-circular pin-hole.Chapter 1,background and significance of this project were introduced.The technology of piston non-circular pin-hole precision machining,approach of GMA intelligence boring bar design,model and control methods of hysteresis nonlinearity were summarized.Then,deficiencies in GMA intelligence boring system at present time were pointed out.Finally,content and significance of this investigation were stated.Chapter 2,the whole model optimization was proceeded and proved based on the prototype.The main components in GMA intelligence boring bar were:magnetic field excitation structure,cooling structure,giant magnetostrictive material(GMM)rod and boring bar.During working,GMM rod extended as the effect of magnetic field produced by excitation structure.The extension of GMM rod lead to bend of boring bar and displacement was exported.As complication coupling among electricity,magnetism,machine and heat,the optimization model from current input to displacement output was difficult to build.In this paper,the difficulties in build whole model was solved effectively by combining the first class Piezomagnetic Equation and artificial neutral networks.The model was solved by genetic algorithm and optimizing results were obtained.At last,availability of this method was proved by simulation.Chapter 3,the great scope frequency response model was difficult to build because of the hysteresis nonlinearity in GMA intelligence boring system.In this paper,the hybrid dynamic model was proposed to settle this problem.The system was divided into two parts:linearity element and rate-dependent hysteresis element.The models were built and identified respectively.Then,two models were connected serially to obtain hybrid dynamic model of GMA intelligence boring system.During research,the coupled response of linearity element and rate-dependent hysteresis element were separated successfully by using combination of inverse model and fitting method.The rate-dependent hysteresis model was builded by dynamic weight function and rate-dependent modified PI model.Finally,the availability and advancement of builded models were proved by experiments.These results increased the precision of dynamic hysteresis nonlinear model and enlarge the scope of frequency response.Meanwhile,these results also supplied the basement for investigation on hysteresis nonlinear system control.Chapter 4,the inverse-model ILC algorithm was proposed to control output displacement exactly by GMA intelligence boring system based on the hybrid dynamic model inverse model.This research solved the difficulties in precision control of GMA intelligence boring system,inhanced the convergence speed about twice compared with the classic iterative learning control.Finally,we also proved availability and advancement of proposed algorithm by experiments.Chapter 5,the GMA intelligence boring system piston non-circular pin-hole experiments were carried out based on inverse model ILC algorithm.Cone and ellipse shape pin-hole manufacturer experiments and analysis were conducted separately.Experiment results showed that precision of cone and ellipse shape pin hole reached 0.45 ?m and 0.91 ?m respectively which was twice better than inverse-model open-loop compensation control method,while iterative number both were 3 which was half of classic ILC's.Chapter 6,the main works of this investigation were concluded and the future research is put forward.