Prediction Of Welding Deformation And Optimization Of Process Parameters For Hull Thin Plates Structure

In order to reduce construction costs and ship weight,the lightweight structure has become the theme of modern shipbuilding.High-strength thin plates not only meet the strength and stiffness requirements of construction,but also reduce the consumption of resources and energy.In hull construction,the plate thickness less than 6mm is generally classified as thin plate,and the thin plate structure mainly exists in the deck of the superstructure and part of the outer plate of the hull.The thin plate structure is more prone to instability deformation during welding,which seriously affects the quality of the construction.As the welding deformation is nonlinear,it is difficult to achieve accurate prediction and control of welding deformation by current technical means.This thesis combines numerical simulation and machine learning theory to propose a thin plate welding deformation prediction and process parameter optimization method.The main research elements are as follows.(1)Welding sequence is an important factor affecting the welding deformation and residual stress,for the hull commonly used T-shaped thin plate structure,the establishment of welding finite element model.The use of double-sided angle welding,the establishment of six different welding sequence program,through the temperature,stress distribution and deformation analysis of each program,showing that warpage deformation is the main form of deformation of thin plate structure,and in the symmetrical isotropic welding deformation and stress are relatively small.Finally,the accuracy of the results is verified by designing symmetrical co directional welding experiments.(2)An improved support vector regression(SVR)method based on particle swarm optimization(PSO)and mesh search is proposed to predict welding deformation of thin plate.First,the design of orthogonal experiments of different process parameters on the deformation of T-shaped thin plate,and the experimental data for extreme difference analysis,the results show that the welding voltage,welding current,welding speed and plate thickness on the deformation and residual stresses.Secondly,the design of particle swarm algorithm and grid search for support vector regression of the key parameters of the search method,the establishment of improved support vector regression deformation and stress prediction model.Through experimental simulation,the results show that the improved prediction model outperforms the original support vector regression algorithm in terms of accuracy and computational speed.(3)A welding process parameter optimization method based on multi-objective Gray Wolf algorithm is proposed.The welding deformation and residual stress as the optimization target,the use of multi-objective gray wolf optimization algorithm to obtain the optimal set of welding parameters solution;then the use of entropy weight method modified by the distance method of superior and inferior solutions to evaluate the solution set,extract the optimal combination of parameters,and finally the group of parameters and the results of the actual processing parameters for comparison,the results show that the optimized parameters so that the deformation and residual stress are reduced,thereby verifying the parameter optimization The effectiveness of the parameter optimization.(4)The development of thin plate welding deformation prediction and process parameter optimization system.The overall framework of the system was built,and the main functional modules of the system were designed.Based on the model of deformation prediction and optimization of process parameters,a human-computer interaction system is developed by using C# language.The online prediction of hull plate welding deformation and optimization of process parameters are realized,which provides guidance for process execution.