Study On Magnetorheological Energy Absorber With Restoring Disc Springs

Impact loading is a long-standing technical challenge for mechanical systems and weaponry.The traditional passive absorbers typically use the orifice with the variable cross-sectional area to generate damping force to dissipate impact energy to improve the safety and stability of the system.The passive absorbers can only achieve a good energyabsorption for a particular impact and cannot well solve the impact problems under other conditions.The advent of magnetorheological(MR)technology has provided a completely new solution to study the adaptive absorbers.However,there are still some key problems that have not been solved well when applying the MR energy absorber(MREA)in the field of high-speed impact,mainly focusing on the physical model,design methodology and dynamic model.Firstly,MREA physical models are quasi-steady models that ignore the inertia effect of MRF,or simplified unsteady models,which are difficult to accurately describe the MREA behaviors.Secondly,MREA design methodologies are based on a quasi-steady model,and use the intelligent algorithms to optimize the geometric parameters,which not only has a large amount of calculation and high cost,but also results in some errors.Finally,most of MREA dynamic models are non-parametric models,and these models can easily fall into local optimums due to the structure and parameter settings,so they cannot accurately predict the dynamic behaviors of MREA.Therefore,in this study,a novel unsteady physical model(called NA-BPIM model)is proposed by using the real acceleration of MRF,and then a new design methodology combining logic analysis with numerical calculation is proposed,and a new dynamic model called JADE-SVR model is proposed by using the adaptive differential evolution(JADE)algorithm to optimize the parameters of support vector regression(SVR).The accuracy of NA-BPIM and JADESVR models,as well as the effectiveness of design methodology,are verified using the high-speed impact tests.The main contents of this thesis are summarized as follows:? The overall design of EA based on the MR valve in parallel to disc spring is studied.By analyzing the main characteristics of high-speed impact,the mixed-mode is employed as a working model of the MR valve,and the disc spring is employed as a restoration element of the MR valve owing to its outstanding capability of providing a large force within narrow deformation.Then,the conceptual diagram of the overall structural design of MREA is proposed,its main structures and working principles are described in detail,and the total MREA force is analyzed.?A novel unsteady physical model is proposed by using the real acceleration to study the inertia effect of MRF.According to the rheological behaviors of MRF tested by the Anton Paal Physica MCR,the Bingham-plastic(BP)model is used to describe the constitutive relationship of MRF,and the parameters of the BP model is identified through the experimental data.Then,the governing equation is developed based on the NavierStokes and continuity equations.In particular,the governing equation is a partial differential equation about position and time variables.Combining the partial differential equation with the boundary and initial conditions of the mixed-mode,a definite solution problem for the MRF velocity distribution is formed.Finally,the method of separation of variables is used to solve the definite solution problem,and the velocity distribution of MRF in each region is deduced,and then the MR valve pressure drop and damping force are obtained.?A new design methodology combining logic analysis with numerical calculation is proposed for the MREA based on the MR valve and disc spring.The design methodology first reduces the number of unknown parameters by logical analysis,and then calculates the remaining parameters by mechanical models.When designing an MR valve,not only a physical model but also a magnetic circuit model is needed,so the magnetic of three-stage coils MR valve is analyzed based on the Kirchhoff s law and the magnetic flux conversation rule.Finally,the design methodology is employed to gradually reduce the number of MR valve dimensions,and then the NA-BPIM model and magnetic circuit model are used to obtain the remaining parameters.It is noted that the unsteady physical model is adopted to optimize the MR valve for the first time.Similarly,all the parameters of disc spring are determined by the proposed design methodology and its mechanical model.?The accuracy of the NA-BPIM model and the effectiveness of the design methodology are validated by the high-speed impact tests of MREA.Two identical MREAs are fabricated according to the design parameters,and a high-speed drop tower facility is set up to carry out a series of impact tests on the two MREAs which are placed in parallel.In order to compare the NA-BPIM model with quasi-steady and simplified unsteady models,a comprehensive evaluation system including qualitatively and quantitatively analysis is developed from three aspects:(1)the accuracy of MREA peak force,(2)the global agreement of dynamic force curves and(3)the accuracy of the dynamic range between model and experiment results.The result shows that the NANPIM model has the highest accuracy.Besides,the impact test results also show that the MREAs achieve the design objectives,which proves the effectiveness of the design methodology.(5)A new hybrid dynamic model for the MREA under high-speed impact is proposed by using the JADE algorithm to optimize SVR parameters.SVR is a statistical learning method based on VC dimension theory and structural risk minimization.It has unique advantages for nonlinear system modeling,and its parameter settings have an important influence on prediction and generalization abilities.To this end,the JADE algorithm with strong search capability and fast convergence speed is used to search for the optimal SVR parameters,and a hybrid dynamics model of MREA is developed by organically combining the two theories.Comparing the JADE-SVR model with GA-SVR,PSO-SVR and DE-SVR models,it is found that the JADE-SVR model has the fastest convergence speed.Besides,the JADE-SVR model is also validated has the highest prediction accuracy by employing six statistical indexes:MAE,MAPE,RMSE,cov,R2 and NSE.Finally,the above four SVR models are compared with the adaptive neuro fuzzy inference system(ANFIS)model.The results show that the accuracy of SVR models is higher than that of the ANFIS model when predicting the dynamic behaviors of MREA under the high-speed impact.