Research On Welding Distortion Prediction Of Ship Plate Frame Structures Based On Self-learning Database

As an important part of our country’s heavy industry,shipbuilding is of great significance to the country’s vigorous development of marine programs and construction of a maritime power.Welding is used as the main joining technology in shipbuilding.Due to the drastic temperature change and uneven residua stress distribution during the welding process,different types of welding distortions are produced in the structure,which affects the assembly accuracy,and even affects the structural strength and the reliability of the whole ship.If the welding distortion distribution of the structure can be quickly and accurately predicted in the ship design stage,it will provide a basis for ship design optimization and selection of corresponding welding distortion corrective measures,so as to provide guarantee for shipbuilding enterprises to grasp the construction period and rationally allocate production resources.In this paper,through the on-site investigation and measurement of the shipyard,based on the thermal-elastic-plastic finite element method(TEP-FEM),the inherent deformation method(IDM)and the application of the database,a welding distortion prediction method of the ship plate frame structure based on the self-learning database was proposed.By selecting the deck support box,the thick stiffened plate and the large plate frame structure as the research objects.The prediction of welding distortion was studied,and the accuracy and applicability of the method were verified by comparing the prediction results with the actual measurement results.To reduce the welding distortion of different types of structures,the control and straightening effects of the flame heating and optimization of the welding sequence on the welding distortion of the structure were studied respectively.In addition,for the welding distortion control of structures with a large number of welds,the Latin hypercube sampling method was used to arrange the weld samples in sequence,and the influence of different welding sequences on the welding distortion was studied.The specific research results are as follows:(1)The concept of establishing a self-learning database including key welding parameters and the inherent deformation of the weld was expounded,and the method of obtaining TEP-FEM sample points and Measured Experience sample points was shown as an example,and a method of creating a self-learning database that conforms to the actual welding process of the shipyard was proposed.(3)Taking the typical deck support box and thick plate stiffened plate of a multipurpose ship as the research objects,the welding deformation prediction research of the multi-pass welding process was carried out.The results showed that the numerical prediction fitted well with the on-site measurement results,and the prediction accuracy could reach within 20%,which verified the feasibility of the prediction method,and found that the transverse bending distortion was the main factor affecting the welding accuracy of the deck support box and the thick stiffened plate structure.To reduce the welding distortion of the structure,the welding distortion correction effects of the methods of flame heating and optimization welding sequence on the deck support box and the stiffened plate structure were studied respectively.The two methods could effectively alleviate the welding distortion of the structure to 20% and 89.6% of the original value.(4)Selecting the large plate frame structure of a cabin as the research object.The welding distortion during the welding process of the large assembly of the structure was obtained.The calculated value was basically consistent with the actual field measurement distortion results,and the maximum error was less than 4 mm and within 20%.The prediction accuracy met the requirements of the shipyard.Then,the welding sequence samples of weld structures with a large number of welds were created based on the Latin hypercube sampling method,and welding deformation predictions were carried out for 5representative welding sequences.The results showed that,considering the welding of stiffened plates,the optimized welding procedure could reduce the distortion of the plate frame structure a maximum of 11.2%.To sum up,through the research on the welding distortion prediction of typical ship plate frame structures,this paper established the welding distortion prediction method of ship plate frame structures based on the self-learning database conforming to the shipyard’s manufacturing process.The welding distortion distribution law of three typical ship structures was summarized,and provided technical support for the shipyard to obtain the welding distortion distribution,optimize the welding sequence and determine the welding compensation in the design stage.This will improve the shipbuilding accuracy and product competitiveness of shipyards.