Research On Optimized Finite Element Method And Uncertainty For Numerical Analysis In Vehicle

With the development of society and the progress of civilization,the requirements for controlling environmental noise,especially automobile noise,are getting higher and higher.To decrease the experiment cost and development cycle,it is necessary to utilize CAE software for virtual analysis and optimization of noise and vibration in design and trial-production stages,the early stage of vehicle development.In the present,the cores of current commercial CAE software are numerical methods like finite element method,boundary element method and meshless method,which play an important role in vehicle noise and vibration analysis.However,boundary element method and meshless method are complicated in calculation and of low efficiency in dealing with practical acoustic engineering problems.Meanwhile,finite element method has some drawback too,such as high stiffness of finite element model,calculation accuracy easily affected by the quality of the grid and calculation frequency and so on.In addition,parameters of acoustic system may be uncertainty by such factors as technology,assembly,measurement and external conditions.These uncertainties are very small,but may produce large errors in predicting the response of the acoustic system when coupling.Therefore,the improvement of the acoustic numerical method and uncertain analysis of acoustic problems has drawn the huge attention of acoustics researchers in recent years,hoping to improve the accuracy,efficiency and engineering practicability of the acoustic numerical calculation.Finite element method(FEM)is a widely used numerical method in CAE software.Its dispersion error determines the accuracy of results.Therefore,we should improve the FEM to reduce the dispersion error and address the problem of low accuracy in FEM solving uncertain acoustic problem.In this paper,the author optimizes the finite element model with or without damp respectively based on the generalized rules and the adaptive genetic algorithm(AGA),and analysis the uncertainties of structural-acoustic system with finite element-least squares point interpolation method(FE-LSPIM).Then,the author applies these methods into some practical projects.The main research work and innovative achievements of this paper are:(1)In order to solve the problem of low accuracy in FEM analyzing high frequency problem due to the over-stiffness of FEM model,this paper introduces the generalized integration rules(GIR)into three-dimension FEM acoustic models.By redistributed the mass matrix,the author yields the optimal equation of three-dimension acoustic FEM analyzation,and then obtain the optimal integration points with adaptive genetic algorithm(AGA).Applying this method into three-dimensional pipe acoustic model,the author analyzed the relationships between the calculation accuracy with mesh size,frequency and the wave propagation angle.Results show that,as compared with the traditional FEM,the optimized finite element method owns higher accuracy and can apply into wider range of frequency acoustic problem.(2)When optimized the FEM,the internal damping of acoustic field is usually neglected by acoustic researchers.In order to study and control the influence of damping or impedance on the accuracy of finite element model,the author applies the OFEM into three-dimension acoustic model with damp,sight the optimal integration with AGA after obtained the optimal equation of three-dimension acoustic problems with damp.Applying this method into three-dimensional pipe acoustic model and practical vehicle model,the author computed the sound pressure responses of these two model.Results show that,the OFEM can gain more accurate results than traditional FEM in three-dimension model with damp too.(3)To reducing the error in FEM analyzing the uncertain structure-acoustic problem,this paper introduced the FE-LSPIM into uncertain structural-acoustic system problems,then yield a novel stochastic perturbation finite element-least square point interpolation method(SP-FE-LSPIM)by combine the method with stochastic perturbation technique.Moreover,this method is validated by taking the plate of structure-acoustic as the research object.Results shows that,as compared with the stochastic perturbation finite element method(SP-FEM),the SP-FE-LSPIM can obtain results that are more accurate.In addition,modal experiments are carried out in this model and experimental result can match the results form FE-LSPIM.It indicated that this method has good engineering applicability.In this paper,the author did some research about optimizing the FEM and analysis the uncertainties in vehicle acoustic problems,studied a generalized integral rule based optimized FEM,improved the accuracy in FEM solving acoustic problem with damp,and developed a SP-FE-LSPIM to accurately analysis uncertain structure-acoustic problem.Numerical examples and experiment show that these methods can efficiently and accurately predicting the vehicle acoustic problem and own good engineering application prospects.