基于LSTM的焙烧时序数据的质量预测模型

Quality Prediction Model of Bake Time Series Data based on LSTM

焙烧是阳极生产中最重要的一环,焙烧的周期为128小时,焙烧的主要工艺参数是焙烧炉各火道的温度及其温度的变化,在整个焙烧期间对铝用阳极的质量有着至关重要的影响。本文针对焙烧工艺参数与质量数据之间的关系进行建模,应用深度学习技术,采用LSTM算法对焙烧历史数据进行训练,旨在找到工艺与质量两者之间的关系,获得质量预测模型,让企业更好把控焙烧过程中的工艺参数,为焙烧生产过程提供辅助作用。实验结果表明,LSTM在阳极质量预测中有较好的效果。

Bake is the most important part of anode production.The bake cycle is 128 hours.The main process parameters of bake are the temperature of each firing path of the bake furnace and its temperature change.The quality of the anode for aluminum is crucial during the whole firing important influence.This article models the relationship between bake process parameters and quality data,applies deep learning technology,and uses the LSTM algorithm to train roasting history data.The purpose is to find the relationship between the process and quality,and to obtain a quality prediction model which is for enterprises to better control the process parameters in the bake process to provide auxiliary effects for the roasting production process.The experimental results show that LSTM has a better effect on anode quality prediction.