摘要
为解决堆叠降噪自动编码器(SDAE)超参数设置不合理而降低离心泵故障诊断精度的问题,选择了金豺狼优化算法(GJO)来优化SDAE超参数。考虑到GJO算法的性能受猎物逃脱能量影响较大的实际,设计了一种自适应逃脱能量策略,得到了自适应金豺狼优化算法(AGJO)。利用AGJO对SDAE超参数进行优化选取,提出了基于AGJO-SDAE的离心泵故障诊断方法。离心泵典型故障诊断实例结果表明,相比于其它方法,AGJO-SDAE在平均诊断精度最少提高了1.03%,在标准差上最少降低了0.007,在耗时上最少减少了4.87 s;在2 dB、8 dB和14 dB噪声强度下,诊断精度相对衰减率最少分别降低了0.21%、1.01%和0.94%。
In order to solve the problem that the stacked denoise auto encoders(SDAE)with unreasonable superparameters reduces the fault diagnosis accuracy of centrifugal pump,a golden jackal optimization algorithm(GJO)was selected to optimized SDAE superparameters.Considering the fact that the performance of GJO algorithm is greatly affected by the prey escape energy,an adaptive escape energy strategy was designed and the adaptive golden jackal optimization algorithm(AGJO)was obtained.Use AGJO optimize SDAE superparameters,a centrifugal pump fault diagnosis method based on AGJO-SDAE was proposed.Typical examples of centrifugal pump fault diagnosis showed that compared with other methods,the average diagnostic accuracy was at least 1.03%higher,the standard deviation was at least 0.007 lower,and the time spent was at least 4.87 s lower.Meanwhile,under the noise intensity of 2 dB,8 dB and 14 dB,the relative attenuation rates of diagnostic accuracy when compared with other methods were reduced by 0.21%,1.01%and 0.94%,respectively.
作者
张毛焕
张伟杰
徐树山
ZHANG Maohuan;ZHANG Weijie;XU Shushan(College of Mechanical and Electrical Information,Shangqiu College,Shangqiu,Henan 476000,China;Electrical Engineering College,Shangqiu Normal University,Shangqiu,Henan 476000,China)
出处
《计量学报》
北大核心
2025年第5期730-737,共8页
Acta Metrologica Sinica
基金
河南省科技攻关项目(242102220018)
河南省教育部产学合作协同育人项目(202102257007)
河南省一流本科课程(豫教【2022】38696)。
关键词
故障诊断
堆叠降噪自编码器
金豺狼优化算法
自适应逃脱能量
离心泵
fault diagnosis
stacked denoising auto encoders
golden jackal optimization
adaptive escape energy
centrifugal pump
