Skip to content
DSML.KZ
Publications

Talk · International Conference on Computer and Applications · December 22, 2025

Real-Time Multi-Parameter Petroleum Coke Quality Monitoring Using Computer Vision and Integrated Anomaly Detection for Aluminum Smelting Operations

Magzhan Kairanbay, Najma Taqi, Basaveswar Reddy Mora Reddy, Ahmed Bahar

Magzhan Kairanbay
Open via DOI

Abstract

Petroleum coke quality control in aluminum smelting relies on manual sampling and offline laboratory analysis, creating temporal gaps, measurement inconsistency, and delayed defect detection. This paper presents an automated computer vision system for real-time petroleum coke particle characterization integrating size distribution analysis with vision-based moisture estimation through an eight-detector parallel anomaly detection framework. The system employs a three-stage preprocessing pipeline (CLAHE, Laplacian sharpening, median filtering) followed by contour-based particle detection achieving measurements spanning 0.1-50mm. Novel contributions include first integrated multi-parameter framework for petroleum coke in aluminum smelting, experimental vision-based environmental parameter estimation eliminating separate sensor infrastructure, and real-time CPU-only processing enabling inline deployment. Validation demonstrates continuous monitoring capability with thousands of particles characterized per minute, providing immediate feedback for process control while maintaining computational efficiency on standard industrial hardware. The system addresses critical gaps in current petroleum coke quality control, offering practical deployment advantages over manual inspection, single-parameter systems, and deep learning approaches requiring extensive training data.

Industrial Vision Systems and Defect DetectionCoal and Coke Industries ResearchMineral Processing and Grindingpetroleum cokealuminium smeltingparticle size distribution

More by Magzhan Kairanbay