Article: Development of coal quality exploration technique based on convolutional neural network and hyperspectral imaging Journal: International Journal of Oil, Gas and Coal Technology (IJOGCT) 2025 Vol.37 No.2 pp.203 - 234 Abstract: Coal is India's prime energy source, contributing about 60% of total electricity production. Coal India, a major coal-producing public sector unit, has produced record 703.2 million tons of coal during the year 2022–2023. Therefore, this paper proposes an idea of instant prediction of coal quality parameters using hyperspectral imaging and deep neural network. We have collected coal samples from 35 different coal mines of all areas of Western Coalfields Ltd (WCL), and 257 different types of samples have been generated. All 257 coal samples were imaged using camera PIKA NIR 320. The RegNet model was applied to predict coal quality based on moisture, ash, volatile matter, gross calorific value, fixed carbon, and sulphur. The results were validated through chemical analysis results received from the lab. The proposed approach achieved good prediction accuracy, nearly 96% for coal quality parameters. Moisture showed the highest accuracy, 96.09% in quality prediction. [Received: October 25, 2023; Accepted: April 14, 2024] Inderscience Publishers - linking academia, business and industry through research

Title: Development of coal quality exploration technique based on convolutional neural network and hyperspectral imaging

Authors: Swati Hira; Manoj B. Chandak; Devendra Kumar Sakhre; Lalit Kumar Sahoo

Addresses: Shri Ramdeobaba College of Engineering and Management, Ramdeo Tekdi, Gittikhadan, Katol Road, Nagpur, 440 013 (M.S.), India ' Shri Ramdeobaba College of Engineering and Management, Ramdeo Tekdi, Gittikhadan, Katol Road, Nagpur, 440 013 (M.S.), India ' CSIR-Central Institute of Mining and Fuel Research, Nagpur Research Centre, Nagpur, India ' CSIR-Central Institute of Mining and Fuel Research, Nagpur Research Centre, Nagpur, India

Abstract: Coal is India's prime energy source, contributing about 60% of total electricity production. Coal India, a major coal-producing public sector unit, has produced record 703.2 million tons of coal during the year 2022–2023. Therefore, this paper proposes an idea of instant prediction of coal quality parameters using hyperspectral imaging and deep neural network. We have collected coal samples from 35 different coal mines of all areas of Western Coalfields Ltd (WCL), and 257 different types of samples have been generated. All 257 coal samples were imaged using camera PIKA NIR 320. The RegNet model was applied to predict coal quality based on moisture, ash, volatile matter, gross calorific value, fixed carbon, and sulphur. The results were validated through chemical analysis results received from the lab. The proposed approach achieved good prediction accuracy, nearly 96% for coal quality parameters. Moisture showed the highest accuracy, 96.09% in quality prediction. [Received: October 25, 2023; Accepted: April 14, 2024]

Keywords: coal quality parameters; hyperspectral imaging; HSI; deep learning; spectral data; spatial data; PIKA NIR-320.

DOI: 10.1504/IJOGCT.2025.144534

International Journal of Oil, Gas and Coal Technology, 2025 Vol.37 No.2, pp.203 - 234

Received: 20 Oct 2023
Accepted: 14 Apr 2024
Published online: 18 Feb 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Development of coal quality exploration technique based on convolutional neural network and hyperspectral imaging

Keep up-to-date