摘要
针对提取离心泵转速数据成本高、操作困难,获得的先验参数误差较大等问题,提出了一种完全基于数据驱动的离心泵转速识别模型。首先,采用快速傅里叶变换(FFT)得到了振动信号的复频谱,利用有效频段内的频谱计算了等效速度谱;然后,通过计算速度谱的谱峰值与本底能量之间的差值,获取了差值最大值,并将其与当前本底能量进行了对比(若满足条件,其对应的频率即为离心泵的转频);对原始振动信号进行了包络解调得到了包络谱,在包络谱中搜寻具有明显谐波特征的故障频率,并分析了故障频率周围的边带特征,结合故障频率和边带特征,对离心泵进行了故障诊断分析;最后,采用实验获得的实际离心泵振动数据,对上述算法模型的准确性进行了验证。实验结果表明:使用基于数据驱动的离心泵转速识别模型对具有明显内圈故障特征的离心泵冲击信号进行识别,其转频为43.58 Hz,在包络谱中有明显的故障特征频率为214.8 Hz,故障特征系数为4.93,符合轴承内圈故障信号特征,也与实际离心泵故障一致。研究结果表明:采用计算速度谱的谱峰值与本底能量之间差值的方法可以有效排除噪声的干扰,准确地识别设备的转速,可为实际工程中的设备故障诊断提供技术支持。
In response to the problems of high cost and operation difficulties in extracting centrifugal pump speed data,and large errors of the obtained priori parameters,a completely data-driven centrifugal pump speed identification method was proposed.Firstly,the complex spectrum of vibration signals was obtained through fast Fourier transform(FFT)calculation,and then the equivalent speed spectrum was calculated by using the spectrum in the effective frequency band.After that,the difference between the spectral peak of the velocity spectrum and the background energy was calculated to obtain the maximum value of difference and compare it with the current background energy.If the condition was satisfied,the corresponding frequency was the rotational frequency of the centrifugal pump.The envelope spectrum was obtained by demodating the original vibration signal,and the fault frequency with obvious harmonic characteristics was searched in the envelope spectrum,and the sideband characteristics around the fault frequency were analyzed.Combining with the fault frequency and sideband characteristics,the fault diagnosis of the centrifugal pump could be analyzed.Finally,the accuracy of the above algorithm model was verified by the actual centrifugal pump vibration data obtained from the experiment.The experimental results show that for the impact signal of the centrifugal pump with obvious inner ring fault characteristics,the rotation frequency is accurately identified as 43.58 Hz,and in the envelope spectrum,the obvious fault feature frequency is 214.8 Hz and the fault feature coefficient is 4.93,which is in line with the fault signal characteristics of the bearing inner ring and is consistent with the actual centrifugal pump fault.The results show that the method of calculating the difference between the spectral peak of the velocity spectrum and the background energy can effectively eliminate noise interference,accurately identify the equipment speed,and provide technical support for equipment fault diagnosis in actual engineer
作者
王建平
杨磊
张欣瑞
陈慧萍
牛会钊
宋健
周同星
WANG Jian-ping;YANG Lei;ZHANG Xin-rui;CHEN Hui-ping;NIU Hui-zhao;SONG Jian;ZHOU Tong-xing(Oil Delivery Team of Oil and Gas Storage and Transportation Branch of Xinjiang Oilfield Company,Changji 831100,China;Beijing Yadan Petroleum Technology Development Co.,Ltd.,Beijing 102200,China;Power Test&Research Institute,Hefei 230601,China)
出处
《机电工程》
CAS
北大核心
2023年第8期1218-1224,共7页
Journal of Mechanical & Electrical Engineering
基金
国家自然科学基金资助项目(51775156)
安徽省自然科学基金资助项目(1908085QE228)。
关键词
离心泵
转速识别
复频谱
等效速度谱
本底能量
故障特征系数
centrifuge pump
speed identification
frequency spectrum
equivalent velocity spectrum
background energy
fault feature coefficient
