摘要
为了提高天然气净化厂中主风机等动设备运维检修效率,以该厂K1401B主风机为例,提出了基于开放域适应网络(Open Set Domain Adaptation,OSDA)的多重交叉故障预测方法。采集主风机振动数据,并运用小波变换去噪,将模型样本映射到源域,并进行目标域划分,目标域中已知故障类型,通过网络进行特征映射到源域完成解耦,根据预测概率分析当前风机故障类型。结果显示网络结构解耦出了90.9%的概率为角度不对中故障,79%的概率为轴承磨损故障,41.7%的概率为停机故障。结论认为,基于开放域适应网络的方法能利用源域中的故障特征库,准确预测目标域故障类型,确保动设备运转过程的安全运行。
In order to improve the overhaul efficiency of the main fan and other dynamic equipments in the natural gas purification plant,a multiple cross-fault prediction method based on Open Set Domain Adaptation is proposed to analyze the fault type for the K1401B main fan.Collected the main fan vibration data and mapped the model samples to the source domain and target domain division after denoising by wavelet transform,the target domain of the known faults class can be mapped to the source domain through the network features to complete the decoupling,analyzed the current fan failure conditions according to the predicted probability.The results show that the network model decoupled 90.9%probability of angular misalignment failure,79%probability of bearing wear failure and 41.7%probability of downtime failure.It is concluded that the open-domain adaptive network can use the fault feature database in the source domain to the fault types in the target domain to ensure the safe operation of the rotated equipments.
作者
刘晓莉
何健聪
Xiao-li Liu;Jian-cong He(Natural Gas Purification Plant;Hunan Eagle Eye Online Electronic Technology Co.,Ltd.)
出处
《风机技术》
2025年第2期90-96,共7页
Chinese Journal of Turbomachinery
关键词
风机
开放域适应网络
多重交叉故障预测
小波变换
安全运行
Fans
Open-Domain Adaptive Network
Multiple Cross-Fault Prediction
Wavelet Transform
Safe Operation
