基于改进PSO-SVR算法的建筑结构钢板激光弯曲成形预测  

Prediction of Laser Bending of Building Structural Steel Plate Based on Improved PSO-SVR

作　　者：高培云 李峰[2] GAO Peiyun;LI Feng(School of Civil Engineering,Shanxi Vocational University of Engineering Science,Jinzhong Shanxi 030619,China;School of Mechanical Engineering,Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]山西工程科技职业大学建筑工程学院,山西晋中030619 [2]太原理工大学机械工程学院,太原030024

出　　处：《机械设计与研究》2023年第2期158-161,共4页Machine Design And Research

基　　金：山西省教育科学“十三五”规划劳动教育专项课题(HLW-20200)。

摘　　要：为了提高高层建筑结构钢板材的激光弯曲成形质量,通过改进粒子群优化算法(PSO)对向量回归机(SVR)进行优化,并对40CrMnNiMo钢板进行激光弯曲成形加工时的功率与扫描速度关键参数进行组合优化,由此获得满足板材弯曲角的最优组合参数。经过训练后的模型进行预测获得的扫描速度与功率达到了跟样本真实参数一致状态。建立模型能够精确预测功率与扫描速度,从而优化激光弯曲成形阶段的各项工艺参数。提高功率后形成了更大板材弯曲角;提高扫描速度后,获得同样弯曲角需要的功率也逐渐升高。提高板材弯曲角后,加工板材达到了更高的线能量。以最低线能量的工况作为最优条件,再利用模型进行预测,由此获得各弯曲角度对应的最优组合参数。该研究为实际生产过程提供指导价值。In order to improve the quality of laser forming of structural steel plate for high-rise buildings,the optimal combination of the plate bending angle parameters are obtained by adopting an improved particle swarm optimization algorithm(PSO)to optimize the vector regression machine(SVR)and optimizing the power and the scanning speed during the laser forming of 40CrMnNiMo structural steel plates.The scanning speed and power obtained by the trained model for prediction are consistent with the real parameters of the sample.The model can accurately predict the power and scanning speed so as to optimize the process parameters in the forming stage of laser bending.After increasing the power,a larger bending angle is formed.With the increase of scanning speed,the power required to obtain the same bending angle also increases gradually.With the increase of bending angle,a higher line energy can be obtained.Taking the working condition with the lowest linear energy as the optimal condition,and using the model to predict,the optimal combination parameters corresponding to each bending angle can be obtained.This study provides the guidance value for actual production process.

关 键 词：激光弯曲 改进PSO-SVR算法 预测误差 工艺参数 

分 类 号：TG665[金属学及工艺—金属切削加工及机床]