摘要
利用已有历史数据,研究多品种小批量物料生产.首先对数据进行预处理,利用主成分分析,根据综合得分得到排名前6的物料为重点关注物料.然后建立多元线性回归、逐步回归以及BP神经网络3种预测模型.利用这3种预测模型分别对6个重点关注物料需求量进行预测,对比误差率,进而得到BP神经网络为最优模型(平均误差率:2.12%).此方法能为企业安排合理的生产计划,创造更多的收益.
In this paper,the production of multi variety and small batch materials is studied.Using the existing historical data,the data is firstly preprocessed,and using principal component analysis,the top 6 materials according to the comprehensive score are got as the key materials.Then,three prediction models including multiple linear regression,stepwise regression and BP neural network are established.The three prediction models are used to predict the demand of six key materials,and the error rate is compared.The BP neural network is the optimal model(average error rate:2.12%).This method can arrange reasonable production plan for enterprises and create more profits.
作者
王静
丁学利
Wang Jing;Ding Xueli(Fuyang Vocational and Technical College)
出处
《哈尔滨师范大学自然科学学报》
CAS
2023年第2期15-22,共8页
Natural Science Journal of Harbin Normal University
基金
安徽省自然科学重点项目(2022AH052570)
安徽省高校优秀青年人才支持计划重点项目(gxyqZD2020077)
阜阳职业技术学院校级科研项目(2021KYMX08)
关键词
主成分分析
BP神经网络
需求量
误差率
Principal component analysis
BP neural network
Demand
Error rate
