Research On Reliability Optimization Method Of Welding Process And Its Application In Magnesium Alloy

Welding process a systematic work which involves various process elements and needs to meet multiple evaluation criteria.The welding process affects the welding quality of the product,and the selection of welding process parameters usually only has the requirements of scope rather than specific value.Therefore,how to adjust the process parameters to meet the performance index is the key to improve the welding process.In the field of commercial manufacture,the requirement of light weight and environmental protection has promoted the wide application of new welding material.However,the welding process is highly dependent on the previous process experience.The welding process of new materials cannot reference the existing process experience,which leads to reliability problem of the welding process.In order to make the quality of new welding materials meet the design requirements and gain the minimum energy consumption in welding production,the design of process parameter combination must be done.But the process require a lot of time and cost,and the efficiency and precision of existing methods to calculate welding performance index are not high.Facing with the above problems,the welding process failure of new developed materials is analyzed.Take the main characteristic parameters(mechanical property and residual stress)of welding process failure as the research object,and aiming at welding process for Mg-5Gd-3Y magnesium alloy material,a universal welding process optimization and reliability optimization method is proposed.The new method improves the ability of new process design,reduces the period of welding process design,and ensures the quality and reliability of welding products.The main research work includes the following aspects:(1)An optimization strategy combining Kriging model and particle swarm algorithm(Kriging-PSO)is proposed,and optimization of mechanical properties and process parameters of AC-TIG butt welding on Mg-5Gd-3Y magnesium alloy sheet using Kriging model and PSO.Firstly,based on Kriging model,the functional relationship between mechanical properties(tensile strength and microhardness)of welded joints and welding process parameters is established.Then,the accuracy of Kriging model is verified by comparing the value of mechanical property measured by the experiment with the predicted value of Kriging model,according to the Kriging model,the mechanical properties of the points to be measured are accurately predicted,and the welding process parameters are optimized by particle swarm intelligent algorithm.The algorithm optimization results are similar to the experimental optimal values,and the effectiveness of the algorithm is verified.Finally,the quality of the optimized weld is evaluated by conventional analysis to verify the accuracy and efficiency of the proposed optimization method.(2)During welding,differences in chemical composition and thermal cycling of magnesium alloy can lead to uneven transition of mechanical properties of welding parts.And the use of welding fixture makes the welding residual stress larger.Facing these problems,the simulation analysis of welding residual stress of newly materials and the study of improved EGO algorithm based on Kriging agent model are carried out.Under the environment of ANSYS,APDL command is used to establish the welding finite element parametric model,then the evolution process of temperature field and residual stress field of Mg-5Gd-3Y magnesium alloy welding parts was studied,finally the weak link of stress concentration of welding parts is found.In order to determine the reasonable welding process parameter range and accurately predict the residual stress of the point to be measured,a parallel finite element EGO optimization method is proposed.The data set obtained by finite element analysis is used to construct the Kriging approximation model,and the Kriging approximation is updated by improving the EGO optimization.The model uses this method to calculate four numerical examples with multiple local advantages.It shows the advantages of the constructed method in terms of convergence speed and accuracy.Finally,the method is applied to establish the approximation between welding residual stress and process parameters.The model accurately predicts the residual stress,prevents the occurrence of residual stress defects in time,reduces the time of simulation calculation,and minimizes the use of the post-weld stress elimination process.(3)Welding residual stress is an important factor affecting the reliability of welding product design and the reliability of manufacturing process design.Because its function in reliability analysis is a multi-dimensional implicit form.A reliability analysis method(AK-MCS+U)combined with active learning U function,Monte Carlo sampling and Kriging model Then the accuracy and efficiency of the method are verified by two numerical examples with different nonlinear degrees.The AK-MCS+U method was used to construct the reliability analysis model of welding residual stress,and the reliability sensitivity analysis was carried out.The quantitative probability was used to reflect the influence of the distribution parameters of random variables on the reliability of the welding process,which provided the reliability optimization of the welding process.The theoretical basis has a high engineering application value.(4)In the deterministic optimization,the optimal value crosses the constraint boundary as the failure solution due to the volatility of the random variable,and the reliability level is low,and the welding product is prone to defects.Therefore,a reliability optimization model of welding process based on defect probability is established,and a reliability optimization method(KPP-SORA)combining physical planning algorithm,AK-MCS+U method and SORA optimization strategy is proposed.The validity and efficiency of the proposed method are verified by an example.The reliability optimization design of the welding process is analyzed by KPP-SORA method.The variation of the objective function value under different reliability requirements is studied.The reliability optimization model of welding process based on defect probability can effectively control the quality of new welded parts and provide theoretical guidance for the research of welding process reliability.