数据驱动的变压器生产物料智能配送研究
Research on Data Driven Intelligent Distribution Method of Transformer Production Materials
姜任奔 1孙文磊 1张学东 1路程 1常赛科 1李祥2
作者信息
- 1. 新疆大学智能制造现代产业学院,乌鲁木齐 830017
- 2. 新疆特变电工股份有限公司,昌吉 831100
- 折叠
摘要
在传统变压器离散装配车间信息化、数字化、智慧化转型的背景下,生产物料的高效智能配送是亟待解决的问题.提出了一种物料备料、路径决策二阶段的生产物料智能配送方法,将物料配送任务分为生产物料备料阶段及生产物料路径自决策阶段.在生产物料备料阶段,提出了基于Adam优化LSTM循环神经网络的变压器生产物料备料时间预测模型,利用生产物料相关时序性数据将Adam优化后的LSTM循环神经网络模型进行训练,准确可靠地实现了对下一阶段生产物料需求时间的预测.在生产物料配送路径自决策阶段,运用五点三次法改进的A∗算法生成了路径平滑,转向较少的配送路径.最后,以某变压器离散制造车间为对象,验证了生产物料的智能化配送方法.
Abstract
Under the background of the informatization,digitalization and intelligence transformation of the traditional transformer discrete assembly workshop,the efficient and intelligent distribution of production ma-terials is an urgent problem to be solved.In this paper,an intelligent distribution method of production materi-als with two stages of material preparation and path decision is proposed.The material distribution task is di-vided into production material preparation stage and production material path self-decision stage.In the pro-duction material preparation stage,a prediction model of transformer production material preparation time based on Adam optimized LSTM recurrent neural network is proposed.The LSTM recurrent neural network model optimized by Adam is trained by using the time series data related to production materials,and the pre-diction of production material demand time in the next stage is accurately and reliably realized.In the self-de-cision stage of production material distribution path,the improved A∗algorithm with five-point cubic method is used to generate a distribution path with smooth path and less turning.Finally,taking a transformer discrete manufacturing workshop as an object,the intelligent distribution method of production materials is verified.
关键词
数据驱动/LSTM循环神经网络/改进A∗算法Key words
data driven/LSTM recurrent neural network/improved A∗ algorithm引用本文复制引用
基金项目
工业互联网标识解析全要素集成平台项目(TC210A02E)
出版年
2024
NETL
NSTL
万方数据