Study On Riveting Deformation And Its Prediction Of Aeronautical Thin-walled Components

As one of the most significant connecting approaches in aircraft assembly,Riveting will induce very small distortion near the rivet hole in local region.With the increase of the number of rivets,the distortion will be accumulated and lead to the twist of the components.Meanwhile,the flight performance and fatigue life will be influenced dramatically.The higher assembly accuracy,shape accuracy and fatigue life are proposed with the structure requirement of new supersonic stealth aircraft,and the riveting distortion has to be strictly controlled.There are numerous factors inducing riveting deformation,and the relationships among these factors are complex.The accumulation regularity of partial distortion by the riveting under the background of multiple riveting needs to be studied further.So the prediction of riveting distortion and the strategy of distortion control make against the improvement of the aircraft assembly technology of aeronautical thin-walled components In this research,combining with theoretical analysis,numerical calculation,experimental research and intelligent optimization,the local-global principle is adopted,and the mechanism of riveting distortion and the accumulation regularity of multiple riveting are further studied,in order to control the distortion by the adjusting of process parameters.The main contents and innovative points are as follows: 1、Based on the contact state,the two phase of single riveting process analysis method was proposed,and the analytical models of relationship between squeeze force and heading geometries,interference fit and heading geometries,radial expansion displacement and squeeze force/interference fit are established.With the contact state of rivet riveting process,the definition of stage division of riveting was proposed,Based on the theory of Унксов,E.П surface friction distribution,the analytical model of squeeze force with heading geometries was established,considering the nonuniform deformation of heading..With the found of heading dip angle and the volume invariance principle,the relation model of interference fit with heading geometries was established.Ignoring the effect of axial stress on the radial expansion,the analytical model of radial expansion displacement with radial stress was established on the basis of uniform interference and thick tube theory.2、The numerical model of riveting assembly distortion with the sinusoidal wave loading method of squeeze force,considering the springback,was established,and the Mind Evolutionary optimization based Algorithm for BP Neural Network learning was proposed to predicting the riveting distortion of single rivet.With the experiments of the quasi-static compress and the Split Hopkinson Pressure Bars,the Johnson-Cook constitutive models of the aluminum alloys of 7075-T651 and 2A10-T4 were built.And the friction coefficients between the aluminum alloys,the aluminum alloy 2A10-T4 and die steel T8 A were obtained.Based on the engineering commonly settings of riveting parameters,the numerical model of riveting assembly distortion with the sinusoidal wave loading method of squeeze force,considering the spring back after the riveting process,was established.With the analysis of calculating results,the variations of stress,strain,displacement and the plastic boundaries of the thin-walled components were obtained,and the amount of distortion of thin-walled components in axial direction is about an order larger than that in radial direction.The presence of heading dip angle was confirmed.The Mind Evolutionary Optimization based Algorithm for BP Neural Network learning was proposed to predicting the riveting distortion of single rivets Setting the process parameters of riveting force,rivet length and rivet/hole clearance that can be controlled by technicians as input,and the amount of interference fit and the distortion of thin-walled components in axial direction as output,the Mind Evolutionary optimization based Algorithm for BP Neural Network learning was proposed.3、The influence laws of the riveting distortion of thin-walled components in axial direction affected by the parameters of riveting process were obtained.The numerical model of multiple riveting with double,three,four and ten rivets were established,which are token the rivet spacing and riveting sequence as the variables.Through the classification of the process parameters involved in the riveting process,the critical process parameters influencing the riveting distortion were determined as squeeze force,rivet length and rivet/hole clearance,and the research showed: For the minimum distortion of thin-walled components,there are optimal values for the critical process parameters.For multi-riveting,the optimal ratio of rivet spacing and rivet diameter for the minimum distortion of thin-walled components was found.4、The Particle Swarm Optimization based Algorithm for Support Vector Regression learning and the Particle Swarm Optimization were proposed to predicting and Optimizing the riveting distortion of multiple rivets.By taking the riveting sequence as input,while the maximum distortion of thin-walled components in axial direction as output,the prediction model for multirivet riveting deformation of thin-walled component based on Particle Swarm Optimization based Algorithm for Support Vector Regression learning was established.Based on the influence laws of the riveting distortion in axial direction affected by the parameters of riveting process,the prediction model of riveting distortion for double row structure with ten rivets was established.On the basis of the prediction model,which was proved by the experiments,an optimization method for riveting sequence based on Particle Swarm Optimization was put forward to minimizing the maximum distortion of thin-walled component,and controlling the distortion induced by riveting.At last,some experiments were carried out to prove the effectiveness of this method for solving the optimization of riveting sequence.