What is the current state of coal mine intelligent construction? What challenges does it face? How can it advance to a higher level? At the recently held 2025 China International Coal Development Forum, coal industry experts shared their insights and engaged in discussions.

“Coal mine intelligence is a critical step in reshaping the industry’s position and a necessary path to making coal the most reliable cornerstone of the new energy system,” said Liu Feng, Vice President of the China National Coal Association, at the forum. “China has formed intelligent construction models for different types of coal mines, including large-scale modern mines, mines with severe hazards, and small and medium-sized mines with complex geological conditions.”

A recent series of thematic press conferences held by the State Council Information Office on the “High-Quality Completion of the 14th Five-Year Plan” revealed that, compared with 2020, the number of intelligent coal mining and excavation working faces nationwide has increased from 494 to 1,930.

Intelligent Demonstration Coal Mines Are Generally at an Intermediate to Primary Level

“The development of mine intelligence in China has been very rapid and of high quality,” said Ge Shirong, Academician of the Chinese Academy of Engineering and President of Jiangxi University of Science and Technology.

As of the end of the first quarter of this year, over 16,000 fixed positions in mines nationwide have achieved unattended operation, with more than 30 types of robots, totaling 2,640 units, and over 2,000 unmanned vehicles being promoted and applied across various mines.

Liu Feng noted that in key coal mining enterprises, the number of single-shift workers in intelligent coal mining faces has decreased by more than six, while labor productivity has increased by over 20%.

Yujialiang Coal Mine of China Energy’s Shendong Coal Company, No. 1 Coal Mine of Huangling Mining Company under Shaanxi Coal Group, and Binhu Coal Mine of Shandong Energy Group have established internationally leading unmanned fully mechanized mining faces, achieving a normal operation model with occasional patrols by personnel during production and unmanned operation.

The unmanned mining truck dispatch system can achieve full coverage of mining equipment, with operational efficiency ranging from 50% to 110% of manual operations. Pure electric unmanned mining trucks have been applied on a large scale in regions such as Xinjiang and Inner Mongolia.

At present, the intelligent demonstration coal mines that have been built remain generally at an intermediate to primary level. In January 2024 and June 2025, the National Energy Administration and the National Mine Safety Administration announced two batches of completed national intelligent demonstration coal mines, totaling 66 mines (including 52 coal preparation plants). Among intelligent underground coal mines, 27 are at the intermediate level of Category I, 31 are at the intermediate level of Category II, and three are at the primary level of Category II. All five intelligent surface coal mines are at the intermediate level. Among intelligent coal preparation plants, 45 are at the intermediate level and seven are at the primary level.

Equipment Support Capability Needs Further Improvement

“Intelligent equipment is still unable to operate fully autonomously,” Ge Shirong pointed out. He identified six major challenges currently facing the normal operation of intelligent mines: insufficient equipment reliability, imprecise perception, poor data interoperability, unstable operation, difficulty in fault prediction, and weak human-machine collaboration.

Qi Yuhao, Deputy Chief Engineer of Shandong Energy Group, also noted that insufficient equipment support capability is one of the challenges facing current coal mine intelligent construction.

Based on practical experience, Qi summarized nine typical issues: first, the foundation of equipment systems is weak, limiting the effectiveness of automated production; second, equipment adaptability is insufficient, making normal operation difficult; third, the failure rate of key components is relatively high, affecting equipment coordination control; fourth, there is a deviation in the direction of industry-university-research application, hindering large-scale promotion; fifth, the focus of safety management has shifted from “people” to “people, machines, networks, and data,” requiring improvements in supporting standards; sixth, the intensity of equipment operation and maintenance has increased, highlighting a shortage of multidisciplinary talent; seventh, the variety of intelligent equipment is extensive, making quality and value assessment difficult; eighth, there are shortcomings in data accumulation and application; and ninth, some key components and technologies still require independent research and development.

In August of this year, the State Council issued the Opinions on Deeply Implementing the “Artificial Intelligence +” Action, promoting the deep integration of artificial intelligence with various industries and fields across the economy and society.

“Current general large language models still represent statistical intelligence and textual intelligence, not true cognitive intelligence,” said Mao Shanjun, Professor at the Institute of Remote Sensing and Geographic Information Systems at Peking University. “Their ability to ‘read ten thousand books’ is strong, but their ability to ‘travel ten thousand miles’ is insufficient, and their application in engineering fields has not met expectations.”

Mao believes that general large language models such as DeepSeek are essentially one-dimensional, making it difficult for them to understand the multidimensional physical world. They lack sufficient professional knowledge and struggle to handle vector-type data such as GIS and CAD.

Currently, the large models applied by coal enterprises are primarily used for processing industry policies, regulations, and public information, lacking deep integration with specific production mine data, operations, and systems, making it difficult to adapt to private, dynamic, and complex mine application scenarios.

In Mao’s view, artificial intelligence large models suitable for coal mines should possess capabilities such as training on vertical domain data, processing multidimensional vector graphics, expressing multidimensional spatiotemporal relationships, understanding rigorous business logic, and enabling multimodal fusion analysis.

Transition from “Single-Point Intelligence” to “Full-System Collaboration”

“Intelligent construction is currently at a critical stage of transitioning from ‘single-point intelligence’ to ‘full-system collaboration,’” Liu Feng stated. He emphasized that the scope of intelligence should extend from production processes to the entire lifecycle of mines, including planning, design, construction, operation, and closure.

Liu believes that evaluation criteria for coal mine intelligence should shift from “whether it exists” to “whether it is used” and “whether it works well,” and from “how much investment is made” to “whether it saves money” and “whether it ensures safety.” The focus should be on quantifiable and verifiable safety and economic benefits.

Ge Shirong emphasized the importance of both safety and economic benefits in coal mine intelligent construction. In his view, advanced intelligent mines should be data-driven, achieving robotization of mining equipment, knowledge-based production data, transparent mining processes, unmanned mining faces, and optimized production decisions. Through digital operation and maintenance, mines can achieve high levels of safety and efficiency.

Ge noted that human-machine collaboration is a core technology that needs to be addressed in the next phase of intelligent construction, and it is also an effective technical approach to solving the other five major challenges facing the normal operation of current intelligent mines. Specifically, four key technologies need to be mastered: digital twin, transparent geology, data intelligence, and embodied intelligence.

To promote intelligent construction, Shandong Energy Group will improve the standard system, establish a reliability certification and access system for intelligent coal mine equipment, eliminate low-quality, high-failure-rate products, create innovation and research and development platforms, promote industry-university-research collaboration, enhance the level of intelligent manufacturing, build “zero-carbon” parks to high standards, strengthen talent support, and form a new operation and maintenance paradigm characterized by “strong equipment and skilled personnel.”

Next, the China National Coal Association will organize the development of operational quality evaluation standards for production systems such as intelligent excavation, as well as auxiliary production systems such as power supply. It will promote the standardization of interface protocols and data formats for intelligent coal machinery equipment, solicit a batch of representative scenarios involving heavy physical labor, publish a list of technical challenges, encourage manufacturers to tackle these challenges through open competition, and promote the application of practical technical equipment and typical practices, accelerating the mechanization or robot replacement of heavy physical labor.

(Source: China Coal News)