薄壁构件GMA增材制造成形特性与尺寸预测OACSTPCD

Geometry character of deposition layers for thin-walled parts in GMA-based additive manufacturing is researched using GMA as the heat source and H08Mn2Si wire as the additive material.Layer geometries are measured by an optical mi-croscope.Geometry characters of deposition layers are analyzed,and its forming mechanism is also illustrated.The results show that the layer geometries in the previous four layers are unstable and have a larger fluctuation.With the increase of the number of deposition layers,the layer height decreases and then reaches a stable value.With the increase of the number of deposition layers,the layer width firstly decreases and then gradually increased to a stable value,and the layer width has a minimum value in the second layer.A series of experiments are carried out for collecting the input-output data by applying a central composite rotatable design.Nonlinear relations between deposition current,travel speed,arc voltage and layer geom-etry are developed based on neural network algorithms.Testing data show that the neural network model has a great predic-tion capability,and maximum relative errors between predicted and measured values are no more than 6.98%.According to the model,closed and open path thin-walled parts are fabricated,demonstrating the effectiveness of the model employed in the adaptive slicing process for thin-walled components.