The chaotic characteristic of the carbon-monoxide utilization ratio in the blast furnace

Dual ensemble online modeling for dynamic estimation of hot metal silicon content in blast furnace system

Hot metal silicon content (HMSC) is usually utilized to measure the quality of hot metal and reflect the thermal status of blast furnace (BF) system. However, most state-of-the-arts ignore the time-varying behavior of BF ironmaking process, which are impractical. Accordingly, a novel dual ensemble online sequential extreme learning machine (DE-OS-ELM) is proposed to establish the online estimation model of HMSC, which can update the data-driven model with the latest operation data. Specifically, an online learning method with recursive modification is first proposed based on OS-ELM (referred to as RM-OS-ELM) to address the modeling with uncertainty. To heel, a dynamic forgetting factor is presented for the dynamic tracking capability enhancement and convergence acceleration. Furthermore, a final updating rule for sequential implementation is constructed by combining the output weights of OS-ELM and RM-OS-ELM based on their corresponding contributions on modeling. Considering the modeling accuracy and curve trend consistency, multiobjective parameter optimization model is also implemented to achieve the satisfactory performance. By taking the proposed DE-OS-ELM, the estimation model of HMSC is established using industrial data. Comprehensive experiments demonstrate that DE-OS-ELM-based HMSC estimation model is more feasible and practical.

Prediction of coal feeding during sintering in a rotary kiln based on statistical learning in the phase space

In alumina rotary kiln production, adjusting the coal feeding rate is the main way to maintain sintering temperature stability during the sintering process, which plays a critical role in improving production quality and reducing energy consumption. In this paper, a novel integrated method (termed PSR-PCA-HMM) is proposed to predict the coal feeding state for optimal control by integrating principal component analysis (PCA) and the hidden Markov model (HMM) based on phase space reconstruction (PSR). First, the thermal signals in rotary kilns are shown to have obvious chaotic characteristics. Second, PSR is utilized to extract the features of the sintering process in a rotary kiln, and PCA is proposed to efficiently reduce the redundancy of the high-dimensional feature space reconstructed by the PSR. Then, considering the nonlinear dynamic characteristic of the sintering process, three HMM models are built to capture the nonlinear dynamic relationship between thermal variables and the corresponding coal feeding state. Finally, the posterior probabilities with respect to the three HMM models are estimated by using the forward algorithm, and the final prediction of coal feeding is determined by the maximized likelihood estimation. Based on field data, the application results indicate that the PSR-PCA-HMM method can significantly improve prediction performance and help realize stable closed-loop control for the sintering temperature.

A Novel Online Sequential Extreme Learning Machine for Gas Utilization Ratio Prediction in Blast Furnaces

Gas utilization ratio (GUR) is an important indicator used to measure the operating status and energy consumption of blast furnaces (BFs). In this paper, we present a soft-sensor approach, i.e., a novel online sequential extreme learning machine (OS-ELM) named DU-OS-ELM, to establish a data-driven model for GUR prediction. In DU-OS-ELM, firstly, the old collected data are discarded gradually and the newly acquired data are given more attention through a novel dynamic forgetting factor (DFF), depending on the estimation errors to enhance the dynamic tracking ability. Furthermore, we develop an updated selection strategy (USS) to judge whether the model needs to be updated with the newly coming data, so that the proposed approach is more in line with the actual production situation. Then, the convergence analysis of the proposed DU-OS-ELM is presented to ensure the estimation of output weight converge to the true value with the new data arriving. Meanwhile, the proposed DU-OS-ELM is applied to build a soft-sensor model to predict GUR. Experimental results demonstrate that the proposed DU-OS-ELM obtains better generalization performance and higher prediction accuracy compared with a number of existing related approaches using the real production data from a BF and the created GUR prediction model can provide an effective guidance for further optimization operation.