A Blast Furnace Throat Temperature Estimation Method Based on Multi-time Scale Online Fusion

Blast furnace (BF) ironmaking is a critical stage in the production of steel. Reasonable operation of BF requires maintaining a stable and effective gas flow, and the throat temperature directly reflects the distribution of gas flow. Due to the harsh environment inside BF, the internal temperature sensors are often damaged. Therefore, estimating the temperature at the BF throat location is crucial. This paper proposes a BF throat temperature estimation method based on multi-timescale online fusion (MTSOF), combining the mechanics of BF operation and a data-driven model. According to the burden movement model, the ore-coke ratio (OCR) is calculated. Subsequently, considering the multi-time scale characteristics within BF, the furnace dataset is divided into a high-frequency dataset and a low-frequency dataset. The support vector regression (SVR) is used to train the low-frequency temperature estimation model, and an integrated gated recurrent unit (I-GRU) network is proposed to train the high-frequency temperature estimation model. Finally, in the testing phase, the proposed online fusion network based on update gate and forget gate integrates the outputs of the low-frequency temperature estimation model and high-frequency temperature estimation model to obtain the final temperature estimation results. This method has been validated on a BF in South China. Compared with the measurement results, the mean absolute percentage error (MAPE), mean absolute error (MAE) and root mean square error (RMSE) of the BF throat temperature estimation results obtained by the proposed method are 2.997%, 1.659°C and 1.978 °C respectively.