基于改进时间序列模型的烧结机机头温度预测  

Temperature prediction of sintering machine head based on improved time series model

作　　者：汪向硕 赵伟 高大伟 张艳兵 闫中华 WANG Xiangshuo;ZHAO Wei;GAO Dawei;ZHANG Yanbing;YAN Zhonghua(Mining Company,Shougang Group Mining Co.,Ltd.,Qian’an 064402,Hebei China)

机构地区：[1]首钢集团有限公司矿业公司,河北迁安064402

出　　处：《烧结球团》2024年第4期44-48,56,共6页Sintering and Pelletizing

摘　　要：针对烧结机机头温度异常变化会严重影响烧结矿质量的问题,本文采用一种基于WD-ARIMAX-GARCH的超短期温度预测模型。由于温度时间序列经常受外部因素的影响,而时间序列模型(ARIMA)只是对历史时间序列的回归,无法体现外部因素的影响,因此该模型加入了外生变量(ARIMAX),并通过引入小波分解(WD),将温度时间序列分解为若干子序列,将各分量进行残差序列检验,对存在异方差特性的分量建立ARIMA-GARCH模型,将该预测结果作为外生变量与分解后的数据重新组合预测。结果表明,基于WD-ARIMAX-GARCH模型的温度预测方法具有较高的精度。Aultra short term temperature prediction model based on WD-ARIMAX-GARCH is adopted to address the serious impact of abnormal temperature changes in the sintering machine head on the quality of sintered ore.Due to the frequent influence of external factors on temperature time series,and the fact that time series models(ARIMA)only regress historical time series and cannot reflect the impact of external factors,exogenous variables(ARIMAX)are added.By introducing wavelet decomposition(WD),the temperature time series is decomposed into several sub sequences,and residual sequence tests are performed on each component.ARIMA-GARCH models are established for components with heteroscedasticity characteristics,and the predicted results are combined with the decomposed data as exogenous variables for re prediction.The results indicate that the temperature prediction method based on WD-ARIMAX-GARCH model has high accuracy.

关 键 词：温度预测 外生变量 ARIMA-GARCH模型 小波分解 

分 类 号：TF046.4[冶金工程—冶金物理化学]