生产计划与调度中的次年产量预测方法  

Output prediction approach of production planning and scheduling in the next year

作　　者：常建涛[1] 仇原鹰[1] 李申[1] 邵晓东[1] 

机构地区：[1]西安电子科技大学电子装备结构设计教育部重点实验室,陕西西安710071

出　　处：《计算机集成制造系统》2013年第7期1648-1654,共7页Computer Integrated Manufacturing Systems

基　　金：陕西省科技攻关资助项目(DF102110401);中央高校基本科研业务费资助项目(K5051204001)~~

摘　　要：针对生产计划与调度中次年产量预测误差较大的问题,提出一种基于动态改进多元线性回归模型的次年产量预测方法。多元线性回归模型将生产管理过程中的制造资源、人力投入、制造工艺、产品报废等生产全周期的影响因素作为建模变量,即将影响次年产量的相关因素尽可能多地包含在预测模型中,从而使模型的预测结果更接近实际产量。运用后推法,将建立的初始多元线性回归模型进行显著性辨别,剔除了对次年产量影响不显著的变量,建立了产量预测的改进多元线性回归模型,在此基础上进一步建立了动态改进多元线性回归模型。将该方法运用到某航空制造企业的次年产量预测中,通过对比模型预测产量和实际产量,证明了该模型在次年产量预测方面的实用性。To overcome the difficulty of output prediction in production planning and scheduling,a novel output prediction approach based on dynamic-improved multiple linear regression model was proposed.The effecting factor of whole production cycle such as manufacturing resources,human input,manufacturing process and product rejection were taken as the modeling variables by multiple linear regression models in production management.Thus the prediction result of model was closer to the actual production.The initial multiple linear regression model was distinguished remarkably by backstepping method,and the low significance parameters for annual output were removed.On this basis,an improved multiple linear regression model was established.By analyzing the dynamic characteristics of this improved model,a dynamic-improved model was obtained.This method was applied to the output prediction in the next year for an aviation manufacturing enterprise.The performance of this model was demonstrated by comparing the prediction output with actual output of the enterprise.

关 键 词：生产计划和调度 次年产量预测 生产管理 多元线性回归 

分 类 号：TH166[机械工程—机械制造及自动化] TP392[自动化与计算机技术—计算机应用技术]