摘要
因丝网印刷产品质量的影响因素众多,且单次实验周期长、实验数据缺乏,本研究以油墨转移率(ITR)为评价指标,以现有数据集作为研究基础,建立了基于凤仙花优化(GBO)算法和最小二乘支持向量回归(LSSVR)的ITR混合预测模型,用GBO算法优化由LSSVR构建的多元数据非线性预测模型的参数。数据集包含102组实验样本,对提出的混合预测模型进行训练和测试。结果表明,ITR混合预测模型的均方根误差为0.0066,平均绝对百分比误差为1.6502%,确定系数为0.8476,优于其他几种基准模型。该预测模型可以在生产过程中,指导经编运动鞋面丝网印花机参数设置。
There are many factors affecting the quality of screen printing products,and the single experiment has a long period and the experimental data are lacking.In this study,with ITR as the evaluation index and the existing data set as the research basis,the ITR hybrid prediction model was established by the GBO algorithm and LSSVR,and it optimized the parameters of the multivariate data non-linear model constructed by LSSVR with the GBO algorithm.The hybrid prediction model was train and tested by the data set including 102 sets of experimental samples.The results showed that RMSE is 0.0066,MAPE is 1.6502%,and R2 is 0.8476 of the ITR hybrid prediction model,which outperformed several other benchmark models.The prediction model can guide the parameter setting of the silk screen printing machine in the production process.
作者
王小辉
李圣普
WANG Xiao-hui;LI Sheng-pu(College of Software,Pingdingshan University,Pingdingshan 467000,China;College of Information Engineering,Pingdingshan University,Pingdingshan 467000,China;College of Information Science and Technology,Donghua University,Shanghai 201620,China)
出处
《数字印刷》
CAS
北大核心
2022年第6期79-84,共6页
Digital Printing
基金
国家重点研发计划“智能机器人”重点专项(No.2018YFB1308800)
工信部智能制造新模式应用项目(No.201746802)。
关键词
凤仙花优化算法
最小二乘支持向量回归
油墨转移率
丝网印刷
Garden Balsam Optimization
Least Squares Support Vector Regression
Ink transfer rate
Screen printing
