基于PBO与CNN的电梯轴承故障检测  

Elevator bearing fault detection based on PBO and CNN

作　　者：吴聪 李梦男 李琨[1] WU Cong;LI Meng-nan;LI Kun(Kunming University of Technology,Kunming 650504,China)

机构地区：[1]昆明理工大学,昆明650504

出　　处：《信息技术》2023年第4期73-78,共6页Information Technology

摘　　要：电梯在生活中已经变得随处可见。而电梯轴承作为重要的部件之一,当出现故障时会造成财产损失,严重的会发生危及生命的事件。卷积神经网络(Convolutional Neural Network,CNN)具有强大的特征学习能力,能够根据电梯轴承在运行的过程中产生的振动信号来检测电梯轴承是否发生故障。文中结合了北极熊算法(Polar Bear Optimization,PBO)对卷积神经网络的参数进行优化。PBO作为自然界的一种启发式优化算法,采用PBO和卷积神经网络相结合的故障诊断方法,对轴承故障识别率达到了99.6%,高于传统的卷积神经网络对电梯轴承的振动产生的特征信号进行故障诊断的识别率。Elevators have become ubiquitous in life.As one of the important components,elevator bearing would cause property loss and life-threatening events when it fails.Convolutional Neural Network(CNN)has strong feature learning ability,which can detect whether the elevator bearing fails according to the vibration signal generated by the elevator bearing during operation.This paper combines Polar Bear Optimization(PBO)to optimize the parameters of convolutional neural network.As a heuristic optimization algorithm in nature,PBO adopts the fault diagnosis method combining PBO and convolutional neural network,and the bearing fault recognition rate reaches 99.6%,which is higher than the recognition rate of traditional convolutional neural network for fault diagnosis of characteristic signals generated by elevator bearing vibration.

关 键 词：电梯 轴承 北极熊算法 卷积神经网络 故障诊断 

分 类 号：TP306.3[自动化与计算机技术—计算机系统结构]