Building “intelligent brains + intelligent factories” for mines

Technological innovation is leading the intelligent transformation of the coal industry. The integration of new information technologies with coal resource development and utilization has accelerated, resulting in the creation of a set of methodologies and solutions that can be easily replicated.

By the end of 2023, the number of intelligent mining faces had increased from more than 400 to almost 1,600. AI and other advanced technologies have led to the development of construction methods that can be used in different coal seam conditions. This has helped us make working conditions safer and boost efficiency.

On July 18, 2023, Shandong Energy Group, Yunding Technology, and Huawei jointly launched the Pangu mine model, the world's first commercial AI model for the energy industry. They had developed an architecture with cloud-edge synergy and thus introduced the "intelligent brain + intelligent factory" mode for the mining industry.

By utilizing algorithm training and closed-loop service management approach, the Pangu model offers comprehensive services for the creation and execution of intelligent scenarios in the energy sector.

In early 2023, Yunding Technology and Huawei agreed to coordinate further R&D into the utilization of the Pangu mine model with the aim of lowering some of the barriers to AI development, enhancing model identification accuracy and generalization, and providing intelligent enablement for safe mine production. That is, the Pangu mine model is able to require only small amounts of data from mine wells for autonomous learning. This can improve safe production across various stages of mining, including digging, drivage, equipment control, transportation, and ventilation.

Safety during production has always been a major point of concern in the coal industry. Drivage (constructing a horizontal or inclined heading or roadway) is a key process. However, traditional operations often take place in confined spaces and involve numerous procedures, making it difficult to ensure both safety and engineering quality.

Yunding Technology achieved a significant breakthrough in intelligent supervision for the drivage process. It figured out how to integrate AI with the drivage processes, technologies, and equipment. Yunding Technology orchestrated on-site construction procedures and developed a complete set of AI scenario applications based on the Pangu mine model. The technology ensures personnel safety, standardizes operation behavior, and guarantees support quality. The Pangu mine model improves the safety and efficiency of coal production, while also making the development of innovative applications more inclusive to the whole industry.

Efficient decision-making is critical in the intelligent transformation of the coal industry. Yunding Technology has developed a refined management solution for coal preparation plants by utilizing the Pangu mine model, and technologies such as IoT, data analysis, and cloud computing. All of these build upon traditional industrial automation. The solution is based on coal preparation plants' automation systems and integrates intelligent operations systems.

For example, in a conventional coking process, the varying compositions of raw coal result in different levels of sulfur and ash. Coal blending is then typically performed manually. This process is heavily dependent on operators' experience, making it difficult to achieve an optimal ratio of sulfur content, ash content, and other characteristics of raw coal. As a result, the quality of coke produced during each coking process can vary.

Yunding Technology tackled this issue by using a prediction model to organize and extract the expertise of its experienced operators and experts. This knowledge was then transformed into a quantifiable and actionable model. By implementing the coking coal blending model, it becomes possible to predict the quality of the resulting coke. Quality prediction, which previously took two to three days, can now be completed in just one minute thanks to the AI.

Pangu models help accumulate knowledge and lower the threshold for AI development. The Pangu mine model combines the experience of experts in both the AI and mining fields. Now, it only takes one or two individuals to develop a single mining application scenario. This scenario can be executed in just two weeks and has been shown to enhance identification accuracy by 20% in real-world testing scenarios. In addition, the workloads required to label data samples can be reduced by more than 80%, freeing up many human resources and allowing for a greater emphasis on developing and operating AI applications.

By 2024 Q1, Yunding Technology had created 67 AI application scenarios across nine key fields, including coal mining, drivage, and transportation, utilizing the visual and predictive capabilities of the Pangu mine model. Now, rapid and extensive replication is taking place. Shandong Energy Group, for example, has implemented 682 scenarios across 32 mining units. Yunding Technology has also kicked off AI project construction for eight coal mining enterprises, including the Wugou Coal Mine of Wanbei Coal-Electricity, and one chemical enterprise.

Pangu models will continue to be used in the mining industry as we explore AI applications in mines to enhance production intelligence. Moreover, Pangu models will extend their influence to other industries including the chemical, logistics, electric power, and new energy industries. This will further promote the extensive application of AI along the entire energy industry and create beneficial outcomes for both upstream and downstream counterparts.