基于改进灰色模型的激光切割粗糙度预测  

Laser cut roughness prediction based on improved gray model

作　　者：葛芸萍[1] 王玉梅[2] GE Yun-ping;WANG Yu-mei(Yellow River Conservancy Technical Institute,Kaifeng 475000,China;Shandong Water Resources Survey and Design Institute,Jinan 250013,China)

机构地区：[1]黄河水利职业技术学院,河南开封475000 [2]山东省水利勘测设计院,山东济南250013

出　　处：《激光与红外》2023年第4期522-527,共6页Laser & Infrared

基　　金：河南省教育厅鉴定项目(No.豫教[2019]30290号)资助。

摘　　要：为了提高激光切割粗糙度预测的精度,提出改进灰色模型。首先基本灰色模型增设线性回归,减少了测试数据的随机波动性;接着通过动态等维灰数递补模型添加新的数据,及时补充和使用原始的信息,每预测一步对参数作一次修正;最后给出了激光切割粗糙度预测流程。试验显示改进灰色模型对激光切割粗糙度预测具有较高的精度,RMSE平均值为0.218,比KF、NN、BPNN、PSO、GM算法减少了32.09%、30.13%、26.10%、21.58%、6.48%,ζ平均值为0.975,比KF、NN、BPNN、PSO、GM算法增加了12.33%、9.43%、7.03%、4.28%、3.07%,评价指标较优。In order to improve the accuracy of laser cut roughness prediction,an improved gray model is proposed.Firstly,linear regression is added to the basic gray model to reduce the random volatility of test data.Secondly,new data is added through the dynamic equal dimensional gray number replenishment model to supplement and used the original information in time,and the parameters are corrected once for each prediction step.Finally,the surface roughness prediction process is given.The experiments show that the improved gray model has higher precision for laser cut roughness prediction,the average value of RMSE is 0.218,which is 32.09%,30.13%,26.10%,21.58%,6.48%less than KF,NN,BPNN,PSO and GM algorithm,ζaverage value is 0.975,which is 12.33%,9.43%,7.03%,4.28%,3.07%more than KF,NN,BPNN,PSO and GM algorithm,and the evaluation index is better.

关 键 词：灰色模型 切割 粗糙度 预测 精度 

分 类 号：TH161[机械工程—机械制造及自动化] TN249[电子电信—物理电子学]