Line Processing Parameters Prediction Of Sheet Metal Forming Base On Artificial Neural Network

With the development of science and technology,ship manufacturing industry gradually farewell labor-intensive industries,digitization,automation,ship plate automatic molding system research is particularly important.Steel plate forming can be divided into hot bending forming and cold bending forming,based on the "automatic forming of ship hull surface plate system" project,puts forward the forming ways of cold and hot working,hot bending forming is using electromagnetic induction heating as heating way,cold bending forming is a pair of concave and convex wheel as a way of cold working.Research train of thought is the use of large deformation elastic finite element calculation from the initial strain distribution needed for the tablet to the target board,then converts it to in-plane strain and the strain of form,according to the distribution of in-plane strain and the strain to determine the appropriate processing parameters and processing path.Therefore,the study of machining parameters and the sheet molding equipment corresponding relationship of strain,outside of the ship plate automatic molding is of great significance.First using ABAQUS finite element software to numerical simulation of the plate forming hot working and cold working on the shaping of strain decomposition conversion,described in the form of in-plane strain and the strain the strain regularity in the process of machining.Then applies PyTorch processing parameters of artificial neural network prediction model is established,through the numerical simulation of a large number of samples,to in-plane strain and the strain as input,processing parameters as output,training and test the artificial neural network model.Finally will be the output of the test sample processing parameters of artificial neural network model in ABAQUS simulation calculation,the calculated strain value and the actual test sample strain value error analysis.Results show that when the "2-30-2" artificial neural network structure model were used respectively to forecast cold and heat working parameters and the "3-7-1" model parameters,forecast the robustness,responsiveness,and convergence are good,after test samples and the actual value of error,to prove that created by the artificial neural network model can well forecast plate molding parameters of hot and cold processing.In order to improve the sample and the efficiency of the debugging of artificial neural network model,developed the plate forming heat and cold processing parameters forecasting system,validated and proved that the system can automatically generated according to the target plate data and artificial neural network forecast model is determined.