新一代人工智能在矿山充填中的应用综述与展望

Research status and perspectives of the application of artificial intelligence in mine backfilling

随着我国浅部矿产资源日趋枯竭,深部资源开采成为矿业发展的必然,而充填采矿法是深部资源开采的主要采矿法之一。目前,充填采矿法的推广受限于充填成本高,如何提高充填设计效率,进而降低充填成本,是矿山充填推广的关键。随着人工智能(AI)技术的发展,以神经网络、决策树等为代表的智能算法正逐渐替代或辅助人类从事各种场景中的简单或复杂工作,推动传统工业领域的跨越式发展。矿业作为传统行业的支柱产业之一,也立足自身发展需要,结合新一代人工智能方法,展开了面向智能化革新的进程。笔者从矿山充填中的人工智能方法出发,简要介绍了人工智能的基本概念及常用的人工智能方法(包括人工神经网络、决策树、随机森林与梯度提升树、支持向量机及非监督式学习方法等)。讨论了人工智能方法在矿山充填中的应用难点,系统分析了新一代人工智能在全尾砂絮凝沉降、充填配比、充填料浆流变及管道输送、充填集成设计及多目标优化等方面的最新研究进展。同时,展望了新一代人工智能在矿山充填中的发展方向(包括性能提升、小数据集问题及应用思路扩展),并提出了智能充填系统的构想。新一代人工智能方法在矿山充填中的发展及应用对实现充填设计的绿色化、智能化和高效化,促进充填技术推广和资源生态协调开采具有重要意义。

With the depletion of near-surface resources in China,deep resource exploitation is the inevitable destiny of the mining industry.Among all mining methods,backfill mining is one of the most important mining methods for deep resource exploitation.The wide application of backfill mining is limited by its high cost.How to increase the efficiency of backfilling design and,thus,reduce backfilling costs is the key to the promotion of mine backfilling.With the development of artificial intelligence(AI)technology,the intelligent algorithms represented by neural networks and decision trees are gradually replacing or assisting humans in simple or complex tasks in various scenarios,which promotes the development of traditional industries.As one of the pillar in traditional industries,the mining industry is trying hard to realize its intellectualization using the state-of-the-art AI methods considering its requirements.Focusing on the application of AI methods in mine backfilling,this paper briefly introduces the basic concepts of AI and the widely-used AI methods(including artificial neural network,decision tree,random forest and gradient boosting machine,support vector machine and some unsupervised learning methods).This paper discusses the application difficulties of AI methods in mine backfilling,and systematically analyzes the research status of the application of AI in the flocculation and sedimentation,mix design,rheological properties and pipe transport of slurry,and integrated design and multi-objective optimization.This paper also presents some future perspectives of the application of AI in mine backfilling(including performance improvement,small datasets problem and the widening of application ideas)and puts forward,for the first time,the concept of intelligent backfilling system.The development and application AI methods in mine backfilling is of great significance to realize green,intelligent and efficient backfilling design,promote the application of backfilling technology and coordinated exploitation of resources and ecology.