Optimization of coal fineness base on tiny data modeling using Gaussian process
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The fineness of coal significantly impacts on the combustion efficiency of boilers and the power consumption of coal mills. The real-time adjustment of the coal fineness via optimized controlling the coal mill is practicable but challenging. The difficulties are about data modeling and the fineness prediction. First, data collection is time-consuming for capturing effective operational conditions which depend on the schedules of a power plant. Only 24 tiny data sets have been collected for 2 months with delicate investigation on which data should be involved from hundreds of collectible values in this work. Second, it is difficult to build an accurate and reliable prediction model on which the following optimization is based using so limited data. This work shows that the prediction ability by using Gaussian process can satisfy the requirements. Before executing optimization step, interpretability analysis is given via a group of experiments in order to confirm that any adjustments of operational parameters followed by the prediction comply with professional knowledge. By setting the prediction as the optimization objective, the operation parameters can be adjusted to the best. This study demonstrates that the data-driven optimization is workable with very limited data and the performance and robustness are convincing as well.
Fields of Study & Subjects
Mineral Processing and Grinding
Gaussian Processes and Bayesian Inference
Fault Detection and Control Systems
Publication details
- Journal / source
- International Journal of Coal Preparation and Utilization
- Pages
- 1–11
- Publication date
- 28 March 2026
- Language
- EN