GM(1，1)模型在铝电解氧化铝浓度估计中的应用研究

Application of gray GM (1,1) model to alumina concentration estimation in aluminum electrolysis

详细介绍了灰色GM(1,1)模型的原理及建立步骤。将GM(1,1)模型引入到铝电解氧化铝浓度估算中,对基于Kalman滤波的氧化铝浓度自适应估算模型进行修正。灰色模型建模维度为5,对采用浓度自适应估算模型得到的浓度与实际浓度的偏差序列进行预测,将预测值反馈到浓度自适应估算模型中,形成新的浓度估算值。通过对电解生产中的30余组氧化铝浓度工艺数据进行估算,结果表明,经修正后,浓度估算值与实际值变化趋势保持一致,基本能满足铝电解工业现场的应用要求。

This paper introduces the theory of GM （ 1,1 ） model and the steps of developing GM model in detail, imports GM （1,1） model into alumina concentration estimation in aluminum electrolysis, and corrects the adaptive model based on Kalman filter theory. The model dimension is 5. We used GM （ 1,1 ） model to predict the difference sequence between the estimation value based on adaptive model and the true value, and sent the GM predicted value into adaptive model to calculate a new alumina concentration estimation value. 30 sets of alumina concentration data were tested. The results show that the predicted average relative error is less than 5% , which satisfies the requirement of aluminum electrolysis production.