Using Machine Learning for Robust Target Prediction in a Basic Oxygen Furnace System

Bae, Jonghee; Li, Yurong; Ståhl, Niclas; Mathiason, Gunnar; Kojola, Niklas

doi:10.1007/s11663-020-01853-5

Cited by 34 publications

(20 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using additional sensors such as these improve prediction accuracy, and may also extend process understanding, but still lacks the transparency to explain influential factors of the collected data. In our previous research, we compared the prediction accuracy between several conventional ML algorithms, trained on a full-variance data set that includes all types of heats from years of production [1]. We found that many earlier publications use limited data sets with a lower variance in the training examples and the target outcomes.…”

Section: Related Workmentioning

confidence: 99%

“…Some conventional ML algorithms are relatively good at explaining model predictions, such as decision trees, k-nearest neighbours. These have been used for temperature prediction in BOF-processes, but with a less satisfactory result [1]. Further, machine learning models that are often considered as easy to interpret, such as linear regression and decision trees, may still be impossible for humans to interpret when the data is high dimensional [18].…”

Section: Related Workmentioning

confidence: 99%

“…Using an ML approach allows for many more heat parameters to be used for target predictions, such as we have done in our earlier work [1]. Utilising the information from such high dimensionality data and for better prediction cannot easily be done otherwise.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Using Reinforcement Learning for Generating Polynomial Models to Explain Complex Data

2021

Self Cite

View full text Add to dashboard Cite

Basic oxygen steel making is a complex chemical and physical industrial process that reduces a mix of pig iron and recycled scrap into low-carbon steel. Good understanding of the process and the ability to predict how it will evolve requires long operator experience, but this can be augmented with process target prediction systems. Such systems may use machine learning to learn a model of the process based on a long process history, and have an advantage in that they can make use of vastly more process parameters than operators can comprehend. While it has become less of a challenge to build such prediction systems using machine learning algorithms, actual production implementations are rare. The hidden reasoning of complex prediction model and lack of transparency prevents operator trust, even for models that show high accuracy predictions. To express model behaviour and thereby increasing transparency we develop a reinforcement learning (RL) based agent approach, which task is to generate short polynomials that can explain the model of the process from what it has learnt from process data. The RL agent is rewarded on how well it generates polynomials that can predict the process from a smaller subset of the process parameters. Agent training is done with the REINFORCE algorithm, which enables the sampling of multiple concurrently plausible polynomials. Having multiple polynomials, process developers can evaluate several alternative and plausible explanations, as observed in the historic process data. The presented approach gives both a trained generative model and a set of polynomials that can explain the process. The performance of the polynomials is as good as or better than more complex and less interpretable models. Further, the relative simplicity of the resulting polynomials allows good generalisation to fit new instances of data. The best of the resulting polynomials in our evaluation achieves a better $$R^2$$ R 2 score on the test set in comparison to the other machine learning models evaluated.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Using Reinforcement Learning for Generating Polynomial Models to Explain Complex Data

2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…The end-point phosphorus (P) content is one of the main end-point targets in steelmaking process due to its significant influence on steel products. 1,2) With the increasing demand for low and ultra-low phosphorus steels, many steel plants apply high-efficiency dephosphorization technology into converter. 3,4) However, in view of the fact that most process parameters are correlative with each other, it is difficult to control the end-point phosphorus content in a high accuracy degree, and thus numerous researches have been carried out to realize the precise control of the end-point P content through the mathematical and intelligent models.…”

Section: Introductionmentioning

confidence: 99%

Prediction Model of End-point Phosphorus Content in Consteel Electric Furnace Based on PCA-Extra Tree Model

2021

View full text Add to dashboard Cite

According to the actual industrial data from a Consteel electric furnace, a prediction model based on the principal component analysis (PCA) and extremely randomized trees (Extra Tree model) is proposed for end-point phosphorus content. PCA is used to reduce the dimensionality of the input variable affecting the end-point phosphorus content and eliminate the collinearity among the input variables, and then the data transformed by PCA are used as input data for the established Extra tree model. Compared with other feature pre-processing methods, PCA method can greatly improve the regression prediction performance of the Extra Tree model. Finally, the validation by test set indicates that for the PCA-Extra Tree model, the hit rates of end-point phosphorus content are 98%, 96% and 89% with the prediction error range of ± 0.005%, ± 0.004% and ± 0.003%, respectively. The combined PCA-Extra Tree model has achieved the effective prediction for end-point phosphorus content, and provided a good reference for the end-point control and judgment of Consteel electric furnace.

show abstract

“…The impurities such as phosphorus should be kept under control in the steelmaking process due to their possible adverse effects on the steel quality. [1] Thus, the endpoint phosphorus (P) content is one of the main target values at the end of smelting operation in the steelmaking process due to its significant influence on the steel products, [2][3][4][5] particularly for low phosphorous steels production. To enhance the dephosphorization efficiency, a lot of industrial experiment is carried out to determine the optimal parameters.…”

mentioning

confidence: 99%

Optimization of Dephosphorization Parameter in Consteel Electric Arc Furnace Using Rule Set Model

Chen

Wang

Chen

2021

steel research int.

View full text Add to dashboard Cite

Consteel electric arc furnace, as one typical short-process smelting equipment, is widely applied in many steel plants. Nevertheless, similar to basic oxygen furnace (BOF), the steelmaking process in Consteel electric arc furnace is also a complicated multivariable high-temperature physicochemical process, and the fluctuation of process parameters in Consteel electric arc furnace would make it difficult to stably control the end-point target. The impurities such as phosphorus should be kept under control in the steelmaking process due to their possible adverse effects on the steel quality. [1] Thus, the endpoint phosphorus (P) content is one of the main target values at the end of smelting operation in the steelmaking process due to its significant influence on the steel products, [2][3][4][5] particularly for low phosphorous steels production. To enhance the dephosphorization efficiency, a lot of industrial experiment is carried out to determine the optimal parameters. For instances, Liu et al. [6] investigated the dephosphorization ratio of hot metal with 80 t BOF, and the experiment results showed that the suitable process conditions for obtaining a high dephosphorization ratio in the converter were that the basicity of slag was 1.2-2.0, and the temperature for the intermediate deslagging was 1600-1700 K. Wu et al., [7] through the industrial test of 120 t converter, determine the process of double slag operation in BOF to stabilize the dephosphorization effect. Yang et al. [8] applied the high-temperature laboratorial experiments to investigate the influence of temperature on the dephosphorization of hot metal at the low temperature range of 1300-1450 C with the slag of the low basicity (CaO/ SiO 2 ) of about 1.8. Tian et al. [9] reported that the effects of Fe 3þ content and the basicity of the dephosphorization slag on dephosphorization were greater than those of the decarbonization slag. Yang et al. [10] make a test in 180 t BOF and determine the range of smelting parameters to reduce the end phosphorus less than 0.007%. The reference mentioned earlier is to optimize the dephosphorization process parameter according to the industrial experimental data; however, optimizing the dephosphorization process parameter according to the actual production data is rare. With the improvement of data storage technology and information collection technology, a large number of actual production data can be obtained. [11,12] From these actual production data, we can obtain a lot of valuable information, which is benefit for adjusting the process parameters to improve the dephosphorization ratio. However, in the face of a large amount of data, it is impossible to mine the valuable information just by human. Data mining techniques are mainly used to mine the hidden information, and their patterns justify their importance in decision making. [13] For examples, Pal and Halder [14] utilized the multiobjective optimization technique to address important issue of achieving low phosphorus content along with high-end blow temperat...

show abstract

Using Machine Learning for Robust Target Prediction in a Basic Oxygen Furnace System

Cited by 34 publications

References 31 publications

Using Reinforcement Learning for Generating Polynomial Models to Explain Complex Data

Using Reinforcement Learning for Generating Polynomial Models to Explain Complex Data

Prediction Model of End-point Phosphorus Content in Consteel Electric Furnace Based on PCA-Extra Tree Model

Optimization of Dephosphorization Parameter in Consteel Electric Arc Furnace Using Rule Set Model

Contact Info

Product

Resources

About