泛采样频率下基于VMD-MTSO-XGBoost的滚动轴承故障诊断  

Rolling Bearing Fault Diagnosis Based on VMD-MTSO-XGBoost at Wide Sampling Frequency

作　　者：赵雅强 刘帅 刘少康[1] 刘卫亮 张启亮 刘长良 武英杰 王昕[5] 康佳垚 ZHAO Yaqiang;LIU Shuai;LIU Shaokang;LIU Weiliang;ZHANG Qiliang;LIU Changliang;WU Yingjie;WANG Xin;KANG Jiayao(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,Hebei Province,China;Baoding Key Laboratory of State Detection and Optimization Regulation for Integrated Energy System,Baoding 071003,Hebei Province,China;Hebei Technology Innovation Center of Simulation&Optimized Control for Power Generation,North China Electric Power University,Baoding 071003,Hebei Province,China;School of Automation Engineering,Northeast Electric Power University,Jilin 132012,Jilin Province,China;National Energy Group New Energy Technology Research Institute,Beijing 102209,China)

机构地区：[1]华北电力大学控制与计算机工程学院,河北保定071003 [2]保定市综合能源系统状态检测与优化调控重点实验室,河北保定071003 [3]华北电力大学河北省发电过程仿真与优化控制技术创新中心,河北保定071003 [4]东北电力大学自动化工程学院,吉林省吉林132012 [5]国家能源集团新能源技术研究院有限公司,北京102209

出　　处：《动力工程学报》2024年第11期1712-1722,共11页Journal of Chinese Society of Power Engineering

基　　金：中央高校基本科研业务费资助项目(2023JG005,2023JC010);河北省高等学校科学技术研究资助项目(CXY2023001);河北省省级科技计划资助项目(22567643H)。

摘　　要：针对机械振动监测系统加速度传感器采样频率不一致而带来的滚动轴承故障诊断速度与准确率降低等问题,提出了泛采样频率下基于变分模态分解结合改进金枪鱼群优化算法优化极端梯度提升树(VMD-MTSO-XGBoost)的滚动轴承故障诊断方法。首先,对振动信号进行小波降噪和降采样处理,得到泛采样频率下的降噪信号;利用变分模态分解(VMD)处理泛采样频率下的降噪信号,提取本征模态函数(IMF)分量指标构成故障特征向量。然后,利用Circle混沌映射初始化金枪鱼群优化(TSO)算法种群,增加初始种群的丰富性和多样性;并采用逐维变异方法对最优个体位置进行扰动,提升算法跳出局部最优的能力,增强算法全局探索能力。最后,利用改进金枪鱼群优化(MTSO)算法对极端梯度提升树(XGBoost)参数进行优化,建立滚动轴承故障诊断模型。采用所提出的故障诊断方法对凯斯西储大学公开数据集、德国帕德博恩大学公开数据集和实测数据集进行了验证。结果表明:在泛采样频率下,相比于其他3种模型,所提出的故障诊断方法可以更加高效、准确地识别滚动轴承故障。In order to solve the problems of low speed and low accuracy of rolling bearing fault diagnosis caused by the inconsistent sampling frequency of the accelerometer of mechanical vibration monitoring system,a rolling bearing fault diagnosis method based on variational mode decomposition-multi-strategy tuna swarm optimization-extreme gradient boosting(VMD-MTSO-XGBoost)at wide sampling frequency was proposed.Firstly,the vibration signal was de-noised by wavelet and de-sampled to get the de-noised signal at wide sampling frequency.The de-noised signal at wide sampling frequency was processed by variational mode decomposition(VMD),and the intrinsic mode function(IMF)component index was extracted to form the fault feature vector.Secondly,the Circle chaotic map was used to initialize the tuna swarm optimization(TSO)population to increase the richness and diversity of the initial population.In order to improve the ability of the algorithm to jump out of the local optimum and enhance the ability of the algorithm to explore the whole world,the dimension-by-dimension mutation method was used to disturb the optimal individual position.Finally,the modified tuna swarm optimization(MTSO)algorithm was used to optimize the parameters of extreme gradient boosting(XGBoost),and the rolling bearing fault diagnosis model was established.The proposed fault diagnosis method was validated by the Case Western Reserve University dataset,the German University of Paderborn dataset and the measured dataset.Results show that at wide sampling frequency,the fault diagnosis method presented in this paper can identify rolling bearing faults more efficiently and accurately compared with the other three models.

关 键 词：滚动轴承 故障诊断 泛采样频率 VMD-MTSO-XGBoost 

分 类 号：TK05[动力工程及工程热物理]