基于AM-PSO-BP神经网络的打印路径规划  

Research on printing path planning based on AM-PSO-BP neural network

作　　者：李冰[1] LI Bin(Urumqi Vocational University,Urumqi 830023,Xinjiang,China)

机构地区：[1]乌鲁木齐职业大学,新疆乌鲁木齐830002

出　　处：《模具技术》2024年第1期33-41,共9页Die and Mould Technology

摘　　要：为提高弧焊焊接效果,提出一种基于AM-PSO-BP神经网络的弧焊打印路径规划方法。方法采用基于自适应方差的自适应变异操作(AM)消除粒子群优化算法(PSO)后期迭代效率低的问题,然后利用AM-PSO算法优化BP(back propagation)神经网络的权重和阈值,实现BP神经网络参数的优化;最后将AM-PSO-BP神经网络算法对弧焊打印工艺参数进行预测,获取更准确的弧焊打印工艺参数。仿真结果表明:所提方法可精确预测弧焊打印工艺参数,在该工艺参数下,弧焊打印的六边形柱体、圆柱体、正方体预测值与实测值相差较小,且在误差允许范围内,具有较高的准确性。以上方法可为精确弧焊打印提供依据。In order to improve the welding effect of arc welding,an arc welding printing path planning method based on AM-PSO-BP neural network is proposed.Firstly,adaptive mutation(AM)based on adaptive variance was adopted to solve the problem of low iterationefficiency of particle swarm optimization(PSO)in the later stage.Then,AM-PSO algorithm was utilized to optimize the weight and threshold of BP(back propagation)neural network,so as to optimize the parameters of BP neural network.Finally,AM-PSO-BP neural network algorithm was used to predict the parameters of arc welding printing process,thereby obtaining more accurate parameters of arc welding printing process.The simulation results show that the proposed method can accurately predict the parameters of arc welding printing process.Under the process parameters,the error between the predicted value and the measured value of the hexagonal cylinder,cylinder and cube printed by arc welding is small,and the error is within the allowable range,which has higher accuracy.It shows that the above method can provide a basis for accurate arc welding printing.

关 键 词：弧焊打印 路径规划 PSO算法 自适应变异 BP神经网络 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]