基于改进XGBoost的带式输送机驱动系统健康状态评估研究  

Research on Health Status Evaluation of Belt Conveyor Drive System Based on Improved XGBoost

作　　者：桂彬彬 周建平[1] 万晓静[1] 肖鹿 周志超 GUI Binbin;ZHOU Jianping;WAN Xiaojing;XIAO Lu;ZHOU Zhichao(College of Mechanical Engineering,Xinjiang University,Urumqi 830017,China;Xinjiang Tianchi Energy Co.,Ltd.,Changji 831700,China)

机构地区：[1]新疆大学机械工程学院,乌鲁木齐830017 [2]新疆天池能源有限责任公司,新疆昌吉831700

出　　处：《煤炭技术》2024年第8期230-234,共5页Coal Technology

基　　金：国家自然科学基金项目(72361032);新疆维吾尔自治区天山英才项目(2022TSYCLJ0044);新疆维吾尔自治区天山创新团队项目(2022D14002);新疆天池能源有限公司科技项目(202207140002)。

摘　　要：针对现有的带式输送机驱动系统健康状态评估方法存在数据分布不均衡、评估指标过多存在冗余性,以及采用单一算法寻找全局最优能力较弱等问题,提出一种基于改进粒子群优化算法(IPSO)优化极端梯度提升算法(IPSO-XGBoost)的带式输送机驱动系统驱动系统健康状态评估方法。通过综合相关系数表征带式输送机驱动系统监测参数之间的相关关系,进而选择合适健康状态评估的指标,通过IPSO算法寻找XGBoost模型的最优参数配置,并将最优参数自动输入XGBoost评估模型进行训练,实现了对带式数据驱动系统健康状态评估。通过实验结果表明所提出的模型能够准确评估带式输送机驱动系统的健康状态。Aiming at the problems of imbalanced data distribution,redundant evaluation indicators,and weak ability to search for global optimal solutions using a single algorithm in existing health status evaluation methods for belt conveyor drive systems,a health status evaluation method for belt conveyor drive systems based on the improved particle swarm optimization(IPSO)algorithm and the extreme gradient boost(IPSO-XGBoost)algorithm is proposed.By characterizing the correlation between the monitoring parameters of the belt conveyor drive system through comprehensive correlation coefficients,and selecting indicators suitable for health status evaluation,the optimal parameter configuration of the XGBoost model was found through the IPSO algorithm,and the optimal parameters were automatically input into the XGBoost evaluation model for training.The health status evaluation of the belt data drive system was achieved.Indicating that the proposed model can accurately evaluate the health status of the belt conveyor drive system.

关 键 词：带式输送机 驱动系统 综合相关系数 粒子群优化算法 极端梯度提升算法 健康状态评估 

分 类 号：TD528[矿业工程—矿山机电]