Bagram Geo

Introduction

Mineral resources are considered the foundation of the national economy [1]. Raw materials for economic development are provided by the mineral processing industry. High-quality development in this sector is essential for transforming traditional mining and aiding economic growth [2], [3]. However, in recent years, raw ore quality has declined due to extensive mining. Labor shortages have intensified, and production efficiency gains have stalled. The mining industry faces significant bottlenecks [4], [5]. Moreover, there is an urgent need for innovation in production methods. Intelligence, as a novel solution, is attracting attention from major mines. It signifies the direction of advanced productivity. Meanwhile, the country emphasizes industrial structure transformation. Numerous documents issued by the government promote digital transformations in mining enterprises. They also expedite the establishment of digital, intelligent, and automated mines. Therefore, constructing intelligence concentrators has become imperative [6], [7].

With increasing global resource demand and increasing emphasis on environmental sustainability, the mineral processing industry is facing unprecedented challenges and opportunities [8]. The development of intelligent technology provides new possibilities for solving these challenges [9], [10]. Because it can improve economic efficiency through precise control and optimization of production processes. At present, intelligent technology has been widely used in key fields such as manufacturing, energy industry, transportation and logistics, medical health, agriculture and mining [11], [12], [13], [14], [15], [16], [17]. Through automated control, data analysis, artificial intelligence, Internet of Things, and cloud computing [18], [19], intelligent technologies can significantly enhance production efficiency and promote sustainable development. Although the initial investment and maintenance costs of such digital infrastructure can be substantial, even a small performance improvement (e.g., 1 %) can generate considerable economic returns in high-tonnage mineral processing operations, thereby offsetting the implementation costs over time. Traditional beneficiation methods have obvious limitations in production efficiency, resource utilization and environmental protection, which are difficult to meet the requirements of modern society for high efficiency, environmental protection and sustainable development. Driven by rapid advances in artificial intelligence (AI), big data analytics, Internet of Things, and robotics, the mineral processing industry is rapidly transforming to intelligence.

Traditional manual or simple mechanical ore sorting methods are gradually unable to meet the needs of modern industry because of their low efficiency, poor accuracy and difficulty in processing complex mineral combinations. Intelligent ore sorting technology combines advanced technologies such as high-resolution cameras, X-ray imaging, and laser scanning, and uses AI algorithms to accurately classify minerals and waste rocks, significantly improving sorting efficiency and accuracy [20], [21]. As a large-scale and complex industrial process, the beneficiation process presents the characteristics of grey box model (Because some of its principles are clear, but there are also unknown mechanisms, and the impact of different factors on the indicators is still unclear). There are some problems, such as various variables, large scale and insufficient real-time information. The grinding process is affected by many factors, such as ore feed amount, ore characteristics, grinding concentration, etc., which leads to the difficulty of maintaining stability of the fineness and particle size distribution of the product, and then affects the effect of the downstream process [22]. Advanced algorithm model is used to model and simulate the grinding process, and the best operating parameters are predicted by combining historical data with real-time data, which can realize adaptive control and optimize the grinding effect [23]. For gravity separation, magnetic separation, flotation and other processes that are difficult to accurately describe by mathematical models, traditional manual control methods often lead to problems such as insufficient accuracy and large fluctuation of indexes due to large process disturbances, uncertain dynamic characteristics and coupling effects, and it is difficult to quickly adjust to the process fluctuations being inside the allowed tolerance. The introduction of machine learning algorithm and intelligent control system to build a multi-parameter coupling control mechanism and online detection system can realize the comprehensive control and optimization of the beneficiation process, improve production efficiency and optimize resource utilization, and ensure that the entire beneficiation process is more efficient, environmentally friendly and economically feasible [24], [25].

Intelligent technology will reduce personnel dependence and minimize manual operations through automation and intelligent systems. Can enhance the innovation capability of enterprises and inject new vitality into them. The intelligent process of mineral processing utilizes intelligent technologies including artificial intelligence, and big data analysis. Applications in production include indicator prediction, intelligent control, image recognition, fault diagnosis, and digital twin visualization[26], [27]. These technologies have improved production efficiency, and reduced energy consumption, and environmental pollution [28], [29]. Therefore, intelligent technology contributes to the high-quality development of the mineral processing industry also promotes industry transformation and upgrading. This review adopts a systematic-narrative hybrid approach to comprehensively analyze AI applications in mineral processing. Focusing on publications from 2015 to 2025, we identified and evaluated more than120 peer-reviewed articles from Web of Science databases. This article studies the artificial intelligence algorithms involved in various processes of mineral processing, research progress, and future technological implementation challenges.

In this study, artificial intelligence algorithms were used as the central framework to review the applications and innovations of different intelligent algorithms in grinding, pre-selection, re selection, flotation, magnetic separation, and roasting processes, which is different from existing reviews that only focus on individual processes. This article also uniquely elaborates on the potential for the collaborative development of emerging technologies such as digital twins, sensor fusion, cyber-physical systems, and artificial intelligence applications. It is pointed out that the mixed CFD-DEM-AI model can be used for work in areas where research is lacking, such as reselection and magnetic separation. The direction of intelligent development in mineral processing was also discussed. This article provides a comprehensive overview of mineral processing intelligence with artificial intelligence algorithms at its core, inspiring the intelligent transformation of the mineral processing industry.

Mineral processing procedure

Mineral processing is the process of separating useful minerals from gangue in ores through physical, chemical, and other methods [30], [31], [32]. Furthermore, it provides qualified raw materials for subsequent smelting or other industrial applications. The mineral processing process mainly includes crushing, preselection, grinding, magnetic separation, gravity separation, flotation, roasting and other processes, as shown in Fig. 1. The function of the preselection process is to preliminarily

Intelligent preselection

Sorting is the process of removing low-grade gangue based on differences in ore characteristics. Sorting is typically used for pre-concentration of the feed material, which can lead to increased head grade and improved production efficiency. Sorting initially relied on visual inspection and manual sorting. With the advancement of intelligent technologies, ore identification now utilizes not only physical parameters such as color, transparency, and optical properties but also spectral analysis

Future direction and challenges

The intelligent transformation of mineral processing is unstoppable. The mineral processing will gradually achieve automation and intelligence. The current research work is providing support for achieving these goals. In the pre-selection, sensor-based technology performed better than pure visual models, such as DE-XRT achieving a lead zinc recovery rate of over 95 % in lead-zinc mines. The combination of hyperspectral imaging and machine learning further improves efficiency by reducing data

Conclusion