基于时频分析和残差神经网络的滚动轴承故障诊断方法

陈瑞; 沙云东

doi:10.7643/issn.1672-9242.2026.09.013

PDF(4178 KB)

装备环境工程 ›› 2026, Vol. 23 ›› Issue (9) : 115-124. DOI: 10.7643/issn.1672-9242.2026.09.013

专刊——航空装备结构动力学

基于时频分析和残差神经网络的滚动轴承故障诊断方法

陈瑞, 沙云东^*

作者信息 +

A Rolling Bearing Fault Diagnosis Method Based on Time-frequency Analysis and Residual Neural Networks

CHEN Rui, SHA Yundong^*

Author information +

文章历史 +

摘要

目的针对滚动轴承振动信号故障特征提取困难且易被噪声成分淹没的问题,提出一种基于常数Q变换与残差神经网络相结合的方法。方法首先,采用常数Q变换对原始振动信号进行处理得到时频图,使用重叠采样法和限制对比度自适应直方图均衡化对时频图进行特征增强,批量处理后得到数据集,并划分训练集与测试集。然后使用残差神经网络（Resnet 18）对划分出的训练集进行训练,利用训练后得到的模型对测试集进行处理与分析,得到基于常数Q变换与残差神经网络相结合的滚动轴承故障诊断模型。结果搭建了滚动轴承典型故障特征提取试验台,利用网络公开数据集对常数Q变换与残差神经网络相结合的滚动轴承故障诊断模型进行了方法有效性验证,并将该方法应用于自建试验台。在本文实验设定的工况与数据集条件下,对轴承故障进行了有效诊断,测试集识别准确率、精确率、召回率及F₁分数均达到100%。结论本文所述基于常数Q变换和残差神经网络的滚动轴承故障诊断方法可为一般机械系统滚动轴承故障诊断与特征识别提供有效手段。

Abstract

The work aims to propose a hybrid method combining the constant Q transform (CQT) with a residual neural network (ResNet 18) to address the challenges associated with fault feature extraction from rolling bearing vibration signals, namely the difficulty in identifying fault characteristics and the susceptibility of these signals to being masked by noise. First, the original vibration signals were processed using the constant Q transform to generate a time-frequency representation. Feature enhancement was then applied to this time-frequency representation employing the overlapping sampling method and limited contrast adaptive histogram equalization. After batch processing, a dataset was constructed and divided into a training set and a test set. Next, a ResNet 18 network was trained on the partitioned training set; the trained model was subsequently used to process and analyze the test set, thereby establishing a rolling bearing fault diagnosis model that integrated the constant Q transform and the residual neural network. A dedicated experimental platform for extracting typical rolling bearing fault features was developed. The effectiveness of this hybrid rolling bearing fault diagnosis model was validated using publicly available dataset; the proposed method was then applied to the in-house experimental platform. Under the operating conditions and dataset specified in this study, effective fault diagnosis was achieved for the rolling bearings, with the recognition accuracy, precision, recall, and F1-score for the test set all reaching 100%. In conclusion, the rolling bearing fault diagnosis method based on the constant Q transform and the residual neural network described in this paper can serve as an effective approach for fault diagnosis and feature recognition in rolling bearings across various mechanical systems.

导出引用

陈瑞, 沙云东. 基于时频分析和残差神经网络的滚动轴承故障诊断方法[J]. 装备环境工程. 2026, 23(9): 115-124 https://doi.org/10.7643/issn.1672-9242.2026.09.013

CHEN Rui, SHA Yundong. A Rolling Bearing Fault Diagnosis Method Based on Time-frequency Analysis and Residual Neural Networks[J]. Equipment Environmental Engineering. 2026, 23(9): 115-124 https://doi.org/10.7643/issn.1672-9242.2026.09.013

中图分类号： V216

参考文献

[1] 杨岗, 徐五一, 邓琴, 等. 基于振动信号的滚动轴承复合故障诊断研究综述[J]. 西华大学学报(自然科学版), 2024, 43(1):48-69.
YANG G, XU W Y, DENG Q, et al.A Review of Rolling Bearing Compound Fault Diagnosis Based on Vibration Signals[J]. Journal of Xihua University (Natural Science Edition), 2024, 43(1):48-69.
[2] 王婧, 许志伟, 刘文静, 等. 滚动轴承健康智能监测和故障诊断机制研究综述[J]. 计算机科学与探索, 2024, 18(4):878-898.
WANG J, XU Z W, LIU W J, et al.Review of Research on Rolling Bearing Health Intelligent Monitoring and Fault Diagnosis Mechanism[J]. Journal of Frontiers of Computer Science & Technology, 2024, 18(4):878-898.
[3] 赖荣燊, 闫高强. 基于卷积神经网络的滚动轴承故障诊断研究综述[J]. 机电工程, 2024, 41(2):194-204.
LAI R S, YAN G Q.Review of Rolling Bearing Fault Diagnosis Based on Convolutional Neural Network[J]. Mechanical & Electrical Engineering Magazine, 2024, 41(2):194-204.
[4] 栾孝驰, 李彦徵, 徐石, 等. 基于小波包变换与CEEMDAN的滚动轴承故障诊断方法[J]. 航空动力学报, 2024, 39(5):20220473.
LUAN X C, LI Y Z, XU S, et al.Rolling Bearing Fault Diagnosis Method Based on Wavelet Packet Transform and CEEMDAN[J]. Journal of Aerospace Power, 2024, 39(5):20220473.
[5] 沙云东, 赵宇, 栾孝驰, 等. 基于多参数信息融合筛选的滚动轴承振动信号特征提取与表征方法[J]. 推进技术, 2023, 44(7):2205050.
SHA Y D, ZHAO Y, LUAN X C, et al.Feature Extraction and Characterization of Rolling Bearing Vibration Signal Based on Multi Parameter Information Fusion and Screening[J]. Journal of Propulsion Technology, 2023, 44(7):2205050.
[6] 闫海鹏, 郝新宇, 秦志英. 基于参数优化VMD-小波阈值的轴承振动信号降噪方法[J]. 机电工程, 2024, 41(2):245-252.
YAN H P, HAO X Y, QIN Z Y.Noise Reduction Method of Bearing Vibration Signal Based on Parameter Optimized VMD-Wavelet Thresholding[J]. Mechanical & Electrical Engineering Magazine, 2024, 41(2):245-252.
[7] 施亦非, 黄宇峰, 王锋, 等. 基于参数优化特征模态分解的强背景噪声下滚动轴承故障诊断[J]. 振动与冲击, 2024, 43(21):107-115.
SHI Y F, HUANG Y F, WANG F, et al.Fault Diagnosis of Rolling Bearing under Strong Background Noise Based on POFMD[J]. Journal of Vibration and Shock, 2024, 43(21):107-115.
[8] SAHU D, DEWANGAN R K, MATHARU S P S. Long Short-Term Memory Based Fault Diagnosis of Rolling Element Bearings Using Vibration Signals[J]. Journal of Vibration and Control, 2026, 32(9/10):2116-2127.
[9] YANG J H, WU C J, SHAN Z, et al.Extraction and Enhancement of Unknown Bearing Fault Feature in the Strong Noise under Variable Speed Condition[J]. Measurement Science and Technology, 2021, 32(10):105021.
[10] 郭盼盼, 张文斌, 崔奔, 等. 一种滚动轴承早期微弱故障检测与诊断方法[J]. 航空动力学报, 2025, 40(3):20230443.
GUO P P, ZHANG W B, CUI B, et al.A Method for Early Weak Fault Detection and Diagnosis of Rolling Bearing[J]. Journal of Aerospace Power, 2025, 40(3):20230443.
[11] YAZDANIANASR M, ZIMROZ R, WYŁOMAŃSKA A, et al. Efficient Rolling Element Bearing Fault Diagnostics in the Presence of Highly Leptokurtic Non-Gaussian Noise[J]. Mechanical Systems and Signal Processing, 2026, 247:113852.
[12] YADAV E, CHAWLA V K, NAVDEEP, et al. Fault Diagnosis in Roller Bearings by Using an Advanced Artificial Intelligence-Based Metrology[J]. MAPAN, 2026, 41(1):131-153.
[13] TRA V, KIM J, ALI KHAN S, et al.Bearing Fault Diagnosis under Variable Speed Using Convolutional Neural Networks and the Stochastic Diagonal Levenberg-Marquardt Algorithm[J]. Sensors, 2017, 17(12):2834.
[14] 田野, 陈姚节, 张莉, 等. 频域特征提取联合双流CNN的轴承故障诊断方法[J]. 噪声与振动控制, 2024, 44(4):167-173.
TIAN Y, CHEN Y J, ZHANG L, et al.Bearing Fault Diagnosis Based on Frequency Domain Feature Extraction and Combined Dual-Stream CNN[J]. Noise and Vibration Control, 2024, 44(4):167-173.
[15] ZHAN Z Q, WAN S K, LI X H, et al.Thermal Behavior Prediction of the Spindle-Bearing System Based on the Adaptive Thermal Network Modeling Method[J]. Engineering Applications of Artificial Intelligence, 2025, 145:110199.
[16] 孙建, 刘同干, 杨洪武, 等. 基于改进CNN的变工况滚动轴承故障诊断[J]. 热能动力工程, 2025, 40(2):176-186.
SUN J, LIU T G, YANG H W, et al.Improved CNN- Based Fault Diagnosis of Rolling Bearings under Variable Operating Conditions[J]. Journal of Engineering for Thermal Energy and Power, 2025, 40(2):176-186.
[17] 高淑芝, 陈雪峰, 张义民, 等. 基于CEEMDAN和熵特征的滚动轴承故障诊断[J]. 机械设计与制造, 2025(1):99-102.
GAO S Z, CHEN X F, ZHANG Y M, et al.Fault Diagnosis of Rolling Bearing Based on CEEMDAN and Entropy Features[J]. Machinery Design & Manufacture, 2025(1):99-102.
[18] 沙云东, 赵俊豪, 栾孝驰, 等. 基于阈值参数判决筛选的航空发动机主轴承故障特征提取方法[J]. 航空动力学报, 2025, 40(3):284-295.
SHA Y D, ZHAO J H, LUAN X C, et al.Aircraft Engine Main Bearing Fault Feature Extraction Method Based on Threshold Parameter Decision Screening[J]. Journal of Aerospace Power, 2025, 40(3):284-295.
[19] 张文灏, 栾孝驰, 沙云东, 等. 基于CEEMDAN与多参数特征筛选的滚动轴承故障提取方法[J]. 燃气涡轮试验与研究, 2025, 38(6):98-110.
ZHANG W H, LUAN X C, SHA Y D, et al.Rolling Bearing Fault Extraction Method Based on CEEMDAN and Multi-Parameter Feature Screening[J]. Gas Turbine Experiment and Research, 2025, 38(6):98-110.
[20] KALAY O C.An Optimized 1-D CNN-LSTM Approach for Fault Diagnosis of Rolling Bearings Considering Epistemic Uncertainty[J]. Machines, 2025, 13(7):612.
[21] PARK J, YOO J, KIM T, et al.Spectral Kurtosis Attention Network (SKAN):Synergizing Signal Processing and Deep Learning for Fault Diagnosis of Rolling Element Bearings[J]. Expert Systems with Applications, 2025, 292:128581.
[22] BATOOL S, ALI ABID A, ASIF M, et al.Kronecker Convolutional Feature Pyramid for Fault Diagnosis in Rolling Bearings[J]. Scientific Reports, 2025, 15:20735.
[23] BALAJI M A, VENKATESH S N, SUGUMARAN V, et al.A Comparative Study of Image Representation for Roller Bearing Fault Diagnosis Using Pretrained Networks[J]. Journal of Engineering, 2025, 2025:4707723.
[24] CHENNANA A, MEGHERBI A C, BESSOUS N, et al.Vibration Signal Analysis for Rolling Bearings Faults Diagnosis Based on Deep-Shallow Features Fusion[J]. Scientific Reports, 2025, 15:9270.
[25] BHARATHEEDASAN K, MAITY T, KUMARASWAMIDHAS L A, et al.Enhanced Fault Diagnosis and Remaining Useful Life Prediction of Rolling Bearings Using a Hybrid Multilayer Perceptron and LSTM Network Model[J]. Alexandria Engineering Journal, 2025, 115:355-369.