A review of automatic coal volume measuring technologies

Dou Li, Haoxuan Luo, Lin Wang, Jin Gao, Li ChengSHS 2023 · DOI

Abstract

In coal-fired power plants, timely inventory of coal storage is essential for optimizing stock, improving coal blending operations, and enabling scientific economic dispatch. Traditional manual methods suffer from high error rates and costs, while ultrasonic technology is hindered by dust and time-difference issues. This paper reviews modern coal measuring technologies based on laser scanning and UAV tilt photography. We analyze their installation methods, characteristics, and applications, emphasizing the generation of 3D point clouds for volume calculation. Furthermore, we explore the potential of using 2D images and neural networks for 3D reconstruction to verify coal stocks when raw sensor data is unavailable, offering a significant path toward accurate supervision.

Introduction: The Economic Necessity of Accurate Inventory

In thermal power generation, coal represents the primary production cost, typically accounting for 70–80% of total expenditures. To maintain profitability amidst fluctuating fuel prices, plants must implement "economic coal blending." This strategy relies on precise, real-time data regarding the volume and type of coal available in the yard.

Traditional inventory methods have proven inadequate for modern industrial needs. Manual measurements are not only slow but pose safety risks to personnel. Ultrasonic sensors, once popular, are frequently "blinded" by the dense coal dust prevalent in storage environments. Basic computer vision techniques often lack the point density required for high-precision modeling. Consequently, there is a critical market demand for robust, automatic technologies that can provide reliable data for central dispatchers.

Modern Measurement Methodologies

The industry has shifted toward two primary technological pillars for automatic volume measurement:

1. Laser Scanning Technology

Laser systems utilize 2D scanners paired with position sensors (such as GPS or high-precision encoders) to construct detailed 3D point clouds. These systems are deployed in several configurations:

  • Fixed Systems: Often mounted directly on bucket-wheel reclaimers or stackers to monitor the pile during operation.
  • Walking Robots: Automated units that traverse overhead tracks. These are particularly effective in enclosed yards, as they provide a top-down perspective and remove humans from dusty environments.
  • Portable Units: Handheld devices used for supplemental manual surveys.
Lidar and data collection hardware mounted on a bucket-wheel reclaimer.

2. UAV Tilt Photography

For large, open-air coal yards, Unmanned Aerial Vehicles (UAVs) offer a high-efficiency alternative. Drones capture overlapping high-definition images following pre-planned flight paths. Techniques like Structure from Motion (SfM) reconstruct 3D geometry by identifying and matching feature points across image sequences, while camera calibration ensures spatial accuracy.

Key Insights and Technical Challenges

A significant challenge in fixed-mount laser scanning is the "blind spot" phenomenon. When sensors are mounted on the arm of a reclaimer, they may be unable to see the peak or the far side of high coal piles, leading to volume underestimation.

Scanning geometry showing the coverage of laser beams from different machine positions.

To mitigate these issues, walking robots on maintenance tracks are increasingly viewed as the superior solution for enclosed facilities. They provide a comprehensive view that eliminates the geometric limitations of reclaimer-mounted hardware.

Future Directions: Verification via 2D Images

A forward-thinking proposal in this field is the use of 2D images from standard inventory reports to reconstruct 3D models. Because power plants may be reluctant to share raw coordinate point cloud data with central authorities, using neural networks to "check" reported volumes against visual evidence provides a vital layer of supervision. Future research focusing on neural network models capable of calibrating stock data directly from 2D images will be essential for ensuring that economic dispatch is based on objective reality.

Conclusion

Laser scanning and UAV tilt photography have largely replaced older, less reliable methods in the coal industry. While laser systems currently dominate due to their robustness, the integration of UAVs and the development of image-based verification models represent the next frontier in achieving transparent and accurate coal inventory management.

Summary updated

Abstract

In coal-fired power plants, how to timely inventory coal storage, in order to optimize the coal stock, and improve coal blending operation, and then carry out economic dispatch scientifically and reasonably, has always been an important topic. The traditional way of artificial method has big error, high cost and serious waste of resources. Ultrasonic measurement technology is greatly affected by time difference and dust. Measurement technology based on computer vision requires interpolation to fit the correction model. The above methods all have large measurement errors and have been eliminated by the market. This paper mainly introduces coal measuring technologies based on laser and uav tilt photography which have been researched and applied more recently. Their respective installation methods, characteristics and applicable working occasions are analyzed. They have the common characteristic that the 3d coordinate point cloud data of coal yard surface needs to be solved before the 3d model of coal yard can be generated and the coal storage amount can be calculated. In the case that the 3d coordinate point cloud data of coal yard of power plant cannot be obtained directly, 3d reconstruction can be carried out by using 2d images of coal yard in coal report, or by using neural network model, and then the coal volume can be checked. This research idea is of great significance to realize effective and accurate supervision of coal storage in power plants..
BibTeX
@inproceedings{li2023review,
  author = {Dou Li and Haoxuan Luo and Lin Wang and Jin Gao and Li Cheng},
  title = {{A review of automatic coal volume measuring technologies}},
  booktitle = {SHS},
  year = {2023},
  doi = {10.1051/shsconf/202316601074},
}
Figure