融合混沌反向学习与分数阶微分的矿石细粒度分类算法  

Fine-grained Ore Classification Algorithm Based on Chaotic Reverse Learning and Fractional Differentiation

作　　者：李聪 张训平 李英强 LI Cong;ZHANG Xunping;LI Yingqiang(College of Mathematics and Statistics,Xinyang College,Xinyang 464000,China;Xinyang Hongxin State-owned Capital Operation Group Co.,Ltd.,Xinyang 464000,China)

机构地区：[1]信阳学院数学与统计学院,河南信阳464000 [2]信阳市宏信国有资本运营集团有限公司,河南信阳464000

出　　处：《金属矿山》2025年第2期167-171,共5页Metal Mine

基　　金：河南省高等学校重点科研项目(编号:22B630019)。

摘　　要：矿石细粒度分类有助于提高矿石品位和回收率,降低能耗和环境污染。然而传统的矿石细粒度分类算法通常基于经验模型或统计学习方法,缺乏对矿石颗粒物理特性和动力学行为的深入理解,导致分类效果不理想。因而提出了一种融合混沌反向学习与分数阶微分的矿石细粒度分类算法,该算法首先利用混沌反向学习方法从矿石颗粒的运动轨迹中提取其物理特征(如形状、密度、硬度等);然后使用分数阶微分方程建立矿石颗粒的动力学模型,描述其在分类器中的运动状态;最后根据矿石颗粒的物理特征和动力学状态进行分类。研究表明:该算法不仅能够有效顾及矿石颗粒的非线性、非平稳和多尺度特性,而且能够实现对矿石颗粒的在线、实时和自适应分类,提高了矿石细粒度分类精度和效率。Fine-grained ore classification is helpful to improve ore grade and recovery rate,reduce energy consumption and environmental pollution.However,the traditional fine-grained ore classification algorithms are usually based on empirical models or statistical learning methods,which lack a deep understanding of the physical characteristics and dynamic behavior of ore particles,leading to unsatisfactory classification results.Therefore,a fine-grained ore classification algorithm based on chaotic reverse learning and fractional differentiation is proposed.Firstly,the physical characteristics(such as shape,density,hardness,etc.)are extracted from the motion trajectory of ore particles by using chaotic reverse learning method.Then,the kinetic model of ore particles is established by fractional differential equation to describe their motion state in the classifier.Finally,the ore particles are classified according to their physical characteristics and dynamic state.The results show that the algorithm can not only effectively take into account the nonlinear,non-stationary and multi-scale characteristics of ore particles,but also realize the online,real-time and adaptive classification of ore particles,and improve the precision and efficiency of fine-grained ore classification.

关 键 词：矿石细粒度分类 混沌反向学习 分数阶微分 分类精度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]