Coal Mine Data Analysis using Multi-Agent Assisted Knowledge Map and DRL Model

To improve mining operations in terms of productivity, safety, and decision-making, intelligent mine architecture makes use of cutting-edge technology like knowledge graphs. By visually representing the interconnected data, processes, as well as equipment in the mine on a knowledge graph, immediate analysis and predictive insights are made possible. An adaptable and dynamic environment is created by integrating numerous data sources in this architecture. These sources include operational metrics, geolocation information, and Internet of Things (IoT) sensors. Better judgments, less downtime, and better resource management are the results of the system's ongoing use of AI and machine learning algorithms to hone its knowledge of the mine's operations. An important step toward the digital revolution of the mining sector, the intelligent mine design promotes sustainable practices and optimizes efficiency. Since they collect data in real-time, installed sensors can anticipate when and where problems may arise, making them ideal for predicting system failure and instability in shafts. Mishaps that often arise from an absence of knowledge or exploitation of resources may be better maintained and prevented with the use of this information. Also, self-driving cars and other IoT automated equipment provide vital information about the mined site. In order to plan and carry out their activities properly, large mining firms need this data. Big mining firms use data-based prediction to make well-rounded judgments. A productive situation is made possible by sensors monitoring everything and by systems connected with the IoT ensuring transparent interaction within the environment. A good knowledge graph will detail where information comes from, how it is organized, where it goes, and where it ends up in an organization. Therefore, it serves as an ever-changing guide to the organization's tacit and explicit knowledge. However, in order to know the environment data, which aids in proper knowledge graph development, Deep Reinforcement Learning is used. In addition, the mining sector makes use of ML for predictive analytics by referencing data from knowledge graphs. A random assessment of the pillar's stability using DRL was suggested by Idris as a viable evaluation approach. The Monte Carlo method is used to determine the basic statistical variables while likelihood density function of the rigidity while deformations modulus of the rock mass. The failure is characterized as the limit state when the pillar's peak strength or strain exceeds its peak strain. Using the closed connection created through the trained DRL model, the pillar's reliability index, as well as failure probability is assessed in proportion to the maximum strain it can bear.