基于机器学习框架的氧化铝种分粒度细化爆发时序预测模型

Time series prediction model of crystal seeds particle size attenuation based on machine learning framework

种子分解工序是决定氧化铝产品质量的关键工序,但是当前人们缺乏对晶种分解机理和分解行为的准确本质分析,导致各氧化铝生产企业很难摆脱分解氢氧化铝粒度周期性细化影响,造成过滤效果变差、产能下降、氧化铝产品质量降低。本文利用有监督机器学习方法,建立循环神经网络模型,对种分粒度细化爆发情况进行时间序列预测。经过生产运行数据验证,该模型预测准确度较高,能够为氧化铝生产企业提供种分氢氧化铝细粒子量在未来100天的变化趋势,为生产争取调控时间,减少晶种粒度细化爆发的程度和发生频率。

Seed precipitation process is the key process to determine the product quaity of alumina.However,at present people lack accurate analysis of seed pred pitation mechanism and precipitation behavior。which makes it difcult for alumina enterprises to get rid of the influence of periodic refinement of seed size,resulting in poor filtration elect,reduced production capacity and reduced product quality.Ihn this paper,the supervised machine learning method is used to establish a recurnent neural network model to predict the time series of seed siae attenuation.Though the verifcation of real data,the puediction accuncy of the model is high,which can povide alumina poduction enterprises with the dhange trend of seed size in the next 100 days,so as.to stive for control time for production and reduce the degree and fequency of seed size attenuation.