Research On Multi-Point Forming Springback Prediction Of Shell Plating And Compensation Method

Shell plating processing was a key link in ship construction.At present,shell plating processing was mainly divided into two categories: cold forming and hot forming.Multi-point forming was a special cold forming processing method,which discrete the traditional overall mold into a regular arrangement of the basic body point array,and by controlling the height of the basic body to fit various shapes,so as to realized the flexible processing of the plate.Although the multi-point forming process has become increasingly mature,springback is still an unavoidable forming defect in cold forming.Springback can lead to dimensional deviation of the processed sheet,which affects the forming accuracy and processing efficiency of the outer hull plate.Therefore,to address the problem of springback in multi-point forming,this thesis proposes a springback prediction method based on an improved whale algorithm optimized generalized regression neural network.Machine learning is used to explore the mapping relationship between processing parameters and springback magnitude in multipoint forming.The method of geometric displacement compensation method was used to solve the rebound problem in sheet forming.The improvement of the processing accuracy of multi-point forming was achieved.The research work in the thesis was as follows:(1)Based on the elastic-plasticity theory,the mechanism of springback generation in plate deformation was analyzed.Then a three-dimensional simulation model of multi-point forming was established with a spherical plate as the research object.The material definition,mesh division and constraint setting are also carried out for the model.The processing process of multi-point forming plate was simulated numerically by using finite element analysis.The influence of multi-point forming process parameters on the springback was further investigated.(2)To explore the mapping relationship between process parameters and rebound size,a springback prediction method based on an improved whale algorithm optimized generalized regression neural network was presented.For the problem of insufficient optimization of the algorithm,an improvement strategy of nonlinear convergence factor and adaptive weights was used.Meanwhile,the sheet forming simulation data were used as samples.The GRNN prediction model was built.To improve the prediction accuracy of the GRNN model,the IWOA algorithm was used to find the optimization of the smoothing factor of the GRNN.The IWOA-GRNN springback prediction model was obtained.The accurate prediction of multipoint forming springback was achieved.(3)To address the springback problem in multi-point forming,a reasonable springback compensation direction is determined by combining the characteristics of multi-point forming process.The geometric displacement compensation method method was adopted.The best springback compensation coefficient was obtained.The accurate compensation of rebound was realized.And compared with the traditional empirical compensation method,it was confirmed that the geometric displacement compensation method along the stamping direction can effectively solve the springback problem.(4)Based on the above research,a multi-point forming process management system for hull outer plate was developed.The overall framework and basic functions of the system were designed based on the analysis of user requirements.The multi-point forming process database was constructed.Through the rebound prediction and compensation function in the system,the accurate prediction and efficient compensation of rebound in the multi-point forming process were realized.Finally,the practicality of the system was verified by system application.