摘要
露天矿环境复杂多变,其运输设备油液数据包含大量维度,这些维度之间存在复杂的非线性关系,并且数据本身还可能呈现出稀疏性,即某些特征值在大多数情况下为零或接近零,而关键特征则可能隐藏在这些稀疏的数据之中。这使得传统方法在处理露天矿油液数据时,容易陷入“维度灾难”,难以准确提取关键特征。基于此,提出基于光谱分析仪的露天矿运输设备油液监测方法。通过设计基于Q100光谱分析仪的露天矿运输设备油液数据采集平台,快速准确地获取油液的分析数据。利用基于特征贡献度的油液数据特征抽取方法,从平台存储的大量数据中筛选出最具影响力的特征,克服了高维数据稀疏性与非线性纠缠导致的特征提取难题。应用改进半监督模糊C-均值聚类算法(ISS-FCM)对优化后的特征集展开聚类分析,实现对油液状态的实时监测。实验结果表明,所提方法能够有效识别对油液监测有用的油液数据特征,提高油液监测的精度。
The environment of open-pit mines is complex and variable.The oil fluid data of their transportation equipment contains numerous dimensions,with complex nonlinear relationships between these dimensions.Additionally,the data itself may exhibit sparsity,meaning that certain feature values are zero or close to zero in most cases,while key features may be hidden within this sparse data.This makes traditional methods prone to the"curse of dimensionality"when processing open-pit mine oil fluid data,making it difficult to accurately extract key features.Based on this,an oil fluid monitoring method for open-pit mine transportation equipment based on a spectrum analyzer is proposed.By designing an oil fluid data acquisition platform for open-pit mine transportation equipment based on the Q100 spectrum analyzer,analytical data of oil fluids can be obtained quickly and accurately.A feature extraction method for oil fluid data based on feature contribution is used to screen out the most influential features from the large amountof data stored in the platform,overcoming the difficulties in feature extraction caused by the sparsity of highdimensional data and nonlinear entanglement.The improved semi-supervised fuzzy C-means clustering algorithm(ISS-FCM)is applied to conduct clustering analysis on the optimized feature set,achieving real-time monitoring of oil fluid status.Experimentalresults show that the proposed method can effectively identify oil fluid data features useful for oil fluid monitoring and improve the accuracy of oil fluid monitoring.
作者
薛万安
姬儒东
杨建国
XUE Wan-an;JI Ru-dong;YANG Jian-guo(Guoneng Beidian Shengli Energy Co.,Ltd.,Ordos Inner Mongolia O26000,China;Inner Mongolia University,Hohhot Inner Mongolia O10021,China)
出处
《计算机仿真》
2025年第11期419-424,共6页
Computer Simulation
基金
鄂尔多斯市科技计划计划项目(2022Y219)。
关键词
光谱分析仪
特征筛选
油液数据监测
Q100 Spectrometer
Feature Screening
ISS-FCM Algorithm
Oil Data Monitoring
