基于IPSO-BPNN的超差孔径向增材搅拌摩擦修复接头抗拉强度预测  

Tensile Strength Prediction of Radial-additive Friction Stir Repaired Out-of-tolerance Hole by IPSO-BPNN

作　　者：武芳竹 刘炳宏 李雨亭 李浩然 胡为[1] WU Fangzhu;LIU Binghong;LI Yuting;LI Haoran;HU Wei(National Defense Key Discipline Laboratory of Aeronautical Advanced Connectivity Technology,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学航空先进连接技术重点学科实验室,沈阳110136

出　　处：《机械工程师》2025年第1期32-34,40,共4页Mechanical Engineer

基　　金：航空科学基金资助项目(2020Z048054002)。

摘　　要：利用改进粒子群(Improved Particle Swarm Optimization,IPSO)算法优化了反向传播神经网络(Back Propagation Neural Network,BPNN)的权值和阈值。以旋转速度、下扎速度、停留时间为输入,抗拉强度为输出,构建了3×6×1三层拓扑结构的2024-T4铝合金超差孔径向增材搅拌摩擦修复接头抗拉强度的IPSO-BPNN预测模型。结果表明,优化后的IPSO-BPNN模型具有较高的预测精度和收敛速度,相对预测误差为1.01%。The improved particle swarm optimization(IPSO)algorithm is used to optimize the back propagation neural network(BPNN)weights and thresholds.The IPSO-BPNN tensile strength prediction model for the radial-additive friction stir repairing joints of out-of-tolerance hole of 2024-T4 aluminum alloy with 3×6×1 three layers topology is established with rotating velocity,plunge speed and dwelling time as input and tensile strength as output.The results show that the optimized IPSO-BPNN prediction model has high prediction accuracy and convergence speed,and the relative prediction error is 1.01%.

关 键 词：径向增材搅拌摩擦修复 2024-T4铝合金 IPSO-BPNN预测模型 抗拉强度 

分 类 号：TG44[金属学及工艺—焊接]