融合BP-ANN-MCSA算法的电弧增材工艺参数寻优策略研究

Study on optimization strategy of process parameters of wire-arc addictive manufacturing combined with BP-ANN-MCSA algorithm

为解决电弧熔丝增材过程中工艺参数协同性问题,进行增材工艺参数优化,基于实验与BP神经网络深度学习构建了增材参数与焊道形状的映射关系,建立了增材工艺参数反算模型。将爬山算法进行改进,并与BP神经网络进行深度融合,根据预设电弧增材形貌目标对电弧熔丝增材工艺参数进行寻优,实现了对不同焊道宽高和焊道表面平整度要求下的增材工艺参数优化。最后,通过实验对该模型进行验证,将实验形状参数代入反算模型,将计算所得优化工艺参数与实验预设增材工艺参数进行对比,各项对应参数相对误差的平均绝对值均在5%以内,优化结果有效,证明了该增材工艺参数反算模型及寻优策略的可靠性与实用性。

To solve the problem of synergy of process parameters in the process of wire-arc additive manufacturing and to optimize the addictive process parameters,the mapping relationship of addictive parameters and weld bead shape was established based on the experiments and BP neural network deep learning,and an inverse calculation model of additive process parameters was established.Then,the hill climbing algorithm was improved and combined deeply with BP neural network,according to the preset arc additive morphology target,the wire-arc additive process parameters was optimized,and the optimization of the additive manufacturing process parameters under different weld bead width and height and weld bead surface flatness requirements was realized.Finally,the model was verified by experiment.The shape parameters in experiment were substituted into the inverse calculation model,the optimized process parameters obtained by calculation were compared with the preset addictive manufacturing process parameters in experiment.The average value of the absolute value of the relative error of each parameter is within 5%.The optimization results are valid,which proves the reliability and practicability of the inverse calculation model and optimization strategy of the additive manufacturing process parameters.