摘要
为了解决滚动轴承早期故障信号微弱、特征提取效果不佳,从而影响故障诊断准确性和效率的问题,本文提出了一种结合信号处理与深度神经网络的故障诊断模型。首先,利用霜冰优化算法对变分模态分解的参数进行优化,以获得最佳模态分量;其次,使用麻雀优化算法对CNN-Transformer组合模型的超参数进行调优;最后,将最优模态分量输入优化后的CNN-Transformer模型,以得到故障分类结果。通过美国CWRU轴承数据集验证,实验结果显示,该模型在准确性和稳定性方面相比其他故障诊断模型有显著提升,能够为工业设备的可靠维护提供更精准的故障诊断支持。
To address the issue of weak early fault signals in rolling bearings and the poor feature extraction performance,which affect the accuracy and efficiency of fault diagnosis,this paper proposes a fault diagnosis model that combines signal processing with deep neural networks.First,the frost ice optimization algorithm is used to optimize the parameters of variational mode decomposition to obtain the optimal mode components.The sparrow search algorithm is subsequently utilized to optimize the hyperparameters of the combined CNN-transformer model.Finally,the optimal mode components are fed into the optimized CNN-transformer model to obtain the fault classification results.The experiment on the U.S.CWRU bearing dataset verifies that the proposed model significantly improves accuracy and stability compared to other fault diagnosis models,providing more precise fault diagnosis support for the reliable maintenance of industrial equipment.
作者
杨雄
石宇城
陈儒晖
贺朋飞
YANG Xiong;SHI Yucheng;CHEN Ruhui;HE Pengfei(Department of Computer Engineering,Zhicheng College,Fuzhou University,Fuzhou 350002,China;College of Computer and Data Science,Fuzhou University,Fuzhou 350108,China)
出处
《贵州大学学报(自然科学版)》
2025年第3期44-51,共8页
Journal of Guizhou University:Natural Sciences
基金
福建省财政科研一般资助项目(2023CZ50)
2024年省级大学生创新创业训练计划资助项目(S202413470008)。
