A soft sensor based on case-based reasoning for iron ores flotation

ABSTRACT

Concentrate grade and tailings grade are two vital parameter indexes in a flotation process. To detect the grade succinctly and continuously, a soft sensor based on case-based reasoning (CBR) is proposed. Historic production data is first switched into the form of a case. The case problem includes feed grade, raw ore grade, raw ore ferrous oxide content, raw ore magnetic iron content, target concentrate grade, target tailings grade, dosage of four kinds of reagents; the case solution includes concentrate grade and tailings grade. Simulation result shows that the CBR soft sensor has a higher accuracy and speed in forecasting both concentrate grade and tailings grade when compared with soft sensors supported by other algorithms. The application result in a Chinese iron core dressing mill indicates that the soft sensor presented by this paper causes no damage to people and it can forecast product quality in real-time.

KEYWORDS:

• Soft sensor
• CBR
• flotation
• iron ore
• data-driven

Acknowledgements

We also thank ANSTEEL Group Corporation for offering us experimental environment.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Bojian Liu http://orcid.org/0000-0003-3939-7028

Additional information

Funding

This work was supported in part by the National Natural Science Foundation of China under Grant 61573088, Grant 61573087, and Grant 61433004.