基于改进相似模型的采煤机轴承剩余寿命预测方法  

A method for predicting the remaining useful life of shearer bearings based on improved similarity model

作　　者：李晓昆 耿毅德[2,3] 王宏伟 付翔[1,2] 王然风 LI Xiaokun;GENG Yide;WANG Hongwei;FU Xiang;WANG Ranfeng(College of Mining Engineering,Taiyuan University of Technology,Taiyuan 030024,China;Center of Shanxi Engineering Research for Coal Mine Intelligent Equipment,Taiyuan University of Technology,Taiyuan 030024,China;Postdoctoral Workstation,Shanxi Coking Coal Group Co.,Ltd.,Taiyuan 030024,China;College of Mechanical and Vehicle Engineering,Taiyuan University of Technology,Taiyuan 030024,China;College of Safety and Emergency Management Engineering,Taiyuan University of Technology,Jinzhong 030600,China)

机构地区：[1]太原理工大学矿业工程学院,山西太原030024 [2]太原理工大学山西省煤矿智能装备工程研究中心,山西太原030024 [3]山西焦煤集团有限责任公司博士后工作站,山西太原030024 [4]太原理工大学机械与运载工程学院,山西太原030024 [5]太原理工大学安全与应急管理工程学院,山西晋中030600

出　　处：《工矿自动化》2023年第5期96-103,共8页Journal Of Mine Automation

基　　金：国家自然科学基金项目(52274157);“科技兴蒙”行动重点专项项目(2022EEDSKJXM010);山西省基础研究计划项目(202103021223123);山西省重点研发计划项目(202102100401015);山西省揭榜招标项目(20201101005)。

摘　　要：采煤机轴承退化过程并非简单的线性或指数关系,应分为不同阶段进行分析。而目前的采煤机轴承剩余使用寿命(RUL)预测方法未充分考虑该因素。针对该问题,提出了一种基于改进相似模型的采煤机轴承剩余寿命预测方法。采用通用的相似模型描述设备退化过程,在此基础上通过对均方根聚类分析,将轴承退化过程划分为平稳运行阶段、初始退化阶段和加速退化阶段,借助传统相似模型思路分段计算采煤机轴承的健康状态并拟合得到退化曲线样本库,通过对离线样本库数据和在线采煤机实时数据进行数据预处理和相似性分析,最终得到采煤机轴承RUL。实验结果表明:基于改进相似模型的采煤机轴承RUL预测方法的误差绝对值均值较卷积门控循环单元(ConvGRU)、空间卷积长短时记忆神经网络(ConvLSTM)、卷积神经网络(CNN)、自组织映射神经网络(SOM)、循环神经网络(RNN)、传统相似模型分别降低了30.49%,7.54%,16.98%,24.74%,17.96%,9.49%,可以较好地预测轴承RUL。现场试验结果表明:对采煤机轴承连续监测87 d,轴承健康状态从0.997逐渐下降到0.972,与现场采煤机轴承实际使用情况基本吻合,验证了该方法的有效性。The degradation process of shearer bearings is not a simple linear or exponential relationship.It should be analyzed in different stages.However,the current prediction method for the remaining useful life(RUL)of shearer bearings does not fully consider this factor.In order to solve this problem,a method for predicting the remaining useful life of shearer bearings based on an improved similarity model is proposed.The model uses a universal similarity model to describe the process of equipment degradation.Based on this,through root mean square clustering analysis,the bearing degradation process is divided into the stable operation stage,initial degradation stage,and accelerated degradation stage.With the help of traditional similarity model ideas,the health condition of shearer bearings is calculated by segment.And it is fitted to obtain a degradation curve sample library,Through data preprocessing and similarity analysis on offline sample library data and real-time data of online shearers,the bearing RUL of the shearer is ultimately obtained.The experimental results show that the mean absolute error values of the RUL prediction method for shearer bearings based on improved similarity model are reduced by 30.49%,7.54%,16.98%,24.74%,17.96%and 9.49%respectively,compared to the convolutional gated recurrent unit(ConvGRU),convolutional long short-term memory neural network(ConvLSTM),convolutional neural networks(CNN),self-organizing map(SOM),recurrent neural networks(RNN),and traditional similarity models.The proposed model can effectively predict bearing RUL.The on-site test results show that after continuous monitoring of the bearing of the shearer for 87 days,the health condition of the bearing is gradually reduced from 0.997 to 0.972.The result is basically consistent with the actual use of the bearing of the shearer on site.It verifies the effectiveness of this method.

关 键 词：采煤机轴承 改进相似模型 剩余使用寿命 轴承退化曲线 RUL预测 

分 类 号：TD67[矿业工程—矿山机电]