基于注意力SRU的离散制造过程生产异常分析
Production Anomaly Analysis of Discrete Manufacturing Process Based on Attention SRU
马云骁 1郭宇 1王胜博1
作者信息
- 1. 南京航空航天大学 机电学院,江苏 南京 210016
- 折叠
摘要
为了判断生产异常对生产过程产生何种影响,在对离散制造过程中产生的物联数据进行详细分析的基础上,设计一种基于注意力SRU神经网络的生产任务剩余完工时间预测模型,通过并行化的特征提取过程并结合注意力机制给予不同时刻信息不同权重来对生产任务的延期交付时间进行预测,达到了解量化生产异常对于生产过程影响程度的目的,且有效提高了算法效率和准确率,从而帮助车间工作人员进行后续决策管控工作.
Abstract
In order to judge the impact of production anomalies on production process,a prediction model for the remaining completion time of production tasks based on attention SRU neural network was designed after detailed analysis of the Internet of Things data generated in discrete manufacturing process.Through the parallel feature extraction process and by combination with the attention mechanism,different weights of information at different moments were given to predict the delayed delivery time of production tasks,which achieves the purpose of quantifying the impact of production anomalies on production process,and effectively improves the efficiency and accuracy of the algorithm,so as to help the workshop staff carry out subsequent decision-making control work.
关键词
离散制造过程/制造物联数据/生产异常分析/注意力机制/SRU神经网络Key words
discrete manufacturing process/Internet of Manufacturing Things data/production anomaly analysis/attention mechanism/SRU neural network引用本文复制引用
基金项目
国防基础科研项目(JCKY2018203A001)
国防基础科研项目(JCKY2019204A004)
出版年
2024
国家科技期刊平台
NSTL
万方数据