摘要
针对强背景噪声下旋转矢量(RV)减速器故障诊断困难的问题,提出了一种自适应噪声完备集合经验模态分解(CEEMDAN)结合冠豪猪算法(CPO)优化变分模态分解(VMD)的RV减速器故障诊断方法。首先,利用自适应噪声完备集合经验模态分解对含噪声目标信号进行了降噪分解,得到了一系列固有模态分量(IMF),再根据峭度值原则,选取了目标模态分量;然后,以包络熵为适应性函数,利用CPO算法对变分模态分解中的分解参数K值和α值进行了寻优计算,得到了最后的[K,α]组合,并对VMD进行了最优参数设置;最后,分解后得到一系列本征模态函数分量,对分解后的目标分量进行了重构,再对重构后的目标分量进行包络谱分析并进行了故障诊断,为了验证CEEMDAN-CPO-VMD方法的优越性,进行了实验对比分析。研究结果表明:经CPO重构后的信号信噪比为9.38,均方根误差为0.036,计算时间为36.59 s;利用CEEMDAN-CPO-VMD方法有效地提取了RV减速器的故障特征;对比验证该方法的可行性,使用频谱包络分析得到的结果,有较多的边频干扰,不能有效地定位故障点;同时,对比麻雀搜索算法(SSA)优化的VMD,经SSA算法重构后的信号信噪比为8.57,均方根误差为0.042,计算时间为50.24 s,相比于SSA算法,CPO算法的信噪比结果提高了0.78 dB,均方根误差降低了0.006,迭代时间减少了13.65 s,有了更好的收敛性,验证了CEEMDAN-CPO-VMD法有更好的诊断效果。该研究成果可为强噪声干扰下的RV减速器故障诊断提供参考。
Aiming at the difficulty of diagnosing faults in a rotary vector(RV)reducer under strong background noise,a fault diagnosis method based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)combined with crested porcupine optimizer(CPO)algorithm to optimize variational mode decomposition(VMD)was proposed.Firstly,the complete ensemble empirical mode decomposition with adaptive noise was used to denoise the noisy target signal,a series of intrinsic mode functions(IMF)were obtained.Based on the kurtosis principle,the target mode components were selected.Then,using envelope entropy as the adaptive function,the CPO algorithm was used to optimize the decomposition parameters K and α in the variational mode decomposition,the final[K,α]combination was obtained and the optimal parameters for VMD was set.Finally,a series of intrinsic mode function components were obtained after decomposition,and the decomposed target components were reconstructed.The reconstructed target components were subjected to envelope spectrum analysis for fault diagnosis.In order to verify the superiority of the CEEMDAN-CPO-VMD method,experimental comparative analysis was conducted.The research results show that the signal-to-noise ratio of the signal reconstructed by CPO is 9.38,the root mean square error is 0.036,and the calculation time is 36.59 s.The CEEMDAN-CPO-VMD method is effectively used to extract fault features of RV reducers.Comparing and verifying the feasibility of this method,the results obtained using spectral envelope analysis show that there is a significant amount of sideband interference,it can not effectively locate the fault point.At the same time,comparing with the sparrow search algorithm(SSA)to optimize VMD,the reconstructed signal signal-to-noise ratio of the SSA algorithm is 8.57,the root mean square error is 0.042,and the calculation time is 50.24 s.Comparing with the SSA algorithm,the CPO algorithm improves the signal-to-noise ratio by 0.78 dB,reduces the root mean square error by 0.006,and reduces the iteration time by 13.65 s.It has better convergence and verifies that the CEEMDAN-CPO-VMD method has better diagnostic performance.It provides a reference for the fault diagnosis of RV reducers under strong noise interference.
作者
郭曼
徐建
蔺梦雄
GUO Man;XU Jian;LIN Mengxiong(School of Information Engineering,Xinxiang Vocational and Technical College,Xinxiang 453000,China;Department of Mechanical Engineering,Nanyang Technician College,Nanyang 473000,China;College of Urban Transit and Logistics,Beijing Union University,Beijing 100101,China)
出处
《机电工程》
北大核心
2025年第8期1490-1501,共12页
Journal of Mechanical & Electrical Engineering
基金
河南省教育厅项目(25B520051)。
关键词
旋转矢量减速器
变速器
自适应噪声完备集合经验模态分解
冠豪猪优化算法
变分模态分解
包络熵
故障分类识别方法
rotary vector(RV)reducer
transmission
complete set empirical mode decomposition with adaptive noise(CEEMDAN)
crested porcupine optimization(CPO)algorithm
variational mode decomposition(VMD)
envelope entropy,fault classification and identification method
