针对传统Dirlik方法在机械振动疲劳损伤计算中,对所有频率分量采取均等处理,易掩盖损伤关键频段信息且受无关噪声干扰,导致损伤计算失真、寿命评估偏差的问题,改进Dirlik方法的疲劳损伤计算与寿命评估。方法 以设备特定动作产生的时域振动波形为输入,先将其转化为功率谱密度;再通过注意力机制,依据材料频率响应特性为不同频率分量动态分配权重,突出损伤关键频段并抑制噪声,生成加权功率谱密度;最后结合加权功率谱密度,通过谱矩提取、形状参数计算、等效载荷转换及损伤量化,完成改进Dirlik方法的疲劳损伤计算与寿命评估。结果 改进方法可有效强化损伤关键频段的能量特征,显著抑制无关噪声频段干扰,其疲劳损伤计算精度得到明显提升。结论 基于注意力机制的改进Dirlik方法突破了传统方法频率均等对待的局限,具备良好的自适应性与可靠性,为机械振动监测中的疲劳寿命评估提供了有效技术方案。
Abstract
The work aims to improve fatigue damage calculation and service life assessment of the traditional Dirlik method to deal with the problem that the traditional Dirlik method treats all frequency components equally in the calculation of mechanical vibration fatigue damage, which easily masks the information of frequency bands critical to damage and is affected by irrelevant noise, leading to distorted damage calculation and biased service life evaluation. Firstly, the time-domain vibration waveforms generated by specific equipment actions were taken as input and converted into power spectral density (PSD). Then, through the attention mechanism, weights were dynamically assigned to different frequency components based on the material's frequency response characteristics to highlight the frequency bands critical to damage and suppress noise, thereby generating a weighted power spectral density. Finally, combined with the weighted power spectral density, the fatigue damage calculation and service life evaluation of the improved Dirlik method were completed through spectral moment extraction, shape parameter calculation, equivalent load conversion, and damage quantification. Experimental results showed that the improved method could effectively enhance the energy characteristics of frequency bands critical to damage, significantly suppress interference from irrelevant noise frequency bands, and remarkably improve the accuracy of fatigue damage calculation. The improved Dirlik method based on the attention mechanism breaks through the limitation of the traditional method's equal treatment of all frequencies, exhibits good adaptability and reliability, and provides an effective technical solution for fatigue life evaluation in mechanical vibration monitoring.
关键词
机械振动 /
注意力机制 /
Dirlik方法 /
疲劳损伤 /
寿命评估 /
功率谱密度
Key words
mechanical vibration /
attention mechanism /
Dirlik method /
fatigue damage /
life evaluation /
power spectral density
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 孙卓成. 海洋结构物非高斯随机过程疲劳损伤分析方法[D]. 大连: 大连理工大学, 2023.
SUN Z C.Fatigue Damage Analysis Method for Non-Gaussian Random Processes of Marine Structures[D]. Dalian: Dalian University of Technology, 2023.
[2] 何橼. 动车组转向架撒砂装置振动疲劳研究[D]. 成都: 西南交通大学, 2022.
HE Y.Research on Vibration Fatigue of Sanding Device of EMU[D]. Chengdu: Southwest Jiaotong University, 2022.
[3] 袁盛铭. 基于频域法的转向架附属设备随机振动疲劳寿命评估方法研究[D]. 成都: 西南交通大学, 2022.
YUAN S M.Study on Random Vibration Fatigue Estimation Method of Bogie Accessory Equipments Based on Frequency Domain Method[D]. Chengdu: Southwest Jiaotong University, 2022.
[4] DIRLIK T, BENASCIUTTI D.Dirlik and Tovo-Benasciutti Spectral Methods in Vibration Fatigue: A Review with a Historical Perspective[J]. Metals, 2021, 11(9): 1333.
[5] 何橼, 吴兴文, 邹洪伟, 等. 动车组轴装撒砂装置随机振动疲劳研究[J]. 机车电传动, 2022(5): 64-71.
HE Y, WU X W, ZOU H W, et al.Investigation on Random Vibration Fatigue of EMU Axle-Mounted Sanding Device[J]. Electric Drive for Locomotives, 2022(5): 64-71.
[6] 闫森, 张筱辰, 罗天健, 等. 基于深度子域自适应卷积网络的旋转机械故障诊断[J]. 噪声与振动控制, 2025, 45(6): 117-123.
YAN S, ZHANG X C, LUO T J, et al.Fault Diagnosis of Rotating Machinery Based on Deep Subdomain Adaptation Convolutional Network[J]. Noise and Vibration Control, 2025, 45(6): 117-123.
[7] 孙俊静, 顾幸生. 基于注意力机制多尺度卷积神经网络的轴承故障诊断[J]. 华东理工大学学报(自然科学版), 2024, 50(2): 247-256.
SUN J J, GU X S.Bearing Fault Diagnosis Based on Multi-Scale Convolutional Neural Network of Attention Mechanism[J]. Journal of East China University of Science and Technology, 2024, 50(2): 247-256.
[8] 钱曙杰, 柯尊高, 黄维纲, 等. 基于FFT算法的扶梯振动故障诊断研究[J]. 自动化与仪器仪表, 2023(10): 73-78.
QIAN S J, KE Z G, HUANG W G, et al.Research on Failure Diagnosing by the Escalator Vibration Signal with FFT Algorithm[J]. Automation & Instrumentation, 2023(10): 73-78.
[9] 郭健鑫, 马帅, 凌思睿, 等. 试车振动数据FFT批处理算法[J]. 仪器仪表标准化与计量, 2023(2): 21-24.
GUO J X, MA S, LING S R, et al.FFT Batch Processing Algorithm for Test Vibration Data[J]. Instrument Standardization & Metrology, 2023(2): 21-24.
[10] 范振芳, 黄晋英, 刘思远, 等. 融合ESPRIT和FFT的动叶片振动参数辨识方法研究[J]. 测试技术学报, 2025, 39(4): 373-381.
FAN Z F, HUANG J Y, LIU S Y, et al.Research on Vibration Parameter Identification Method of Rotating Blade Based on ESPRIT and FFT Algorithm[J]. Journal of Test and Measurement Technology, 2025, 39(4): 373-381.
[11] 张明, 李娟, 王浩. 航空薄辐板弧齿锥齿轮行波共振应力响应特性研究[J]. 航空学报, 2024, 45(12): 345-356.
ZHANG M, LI J, WANG H.Study on Stress Response Characteristics of Traveling Wave Resonance for Aero Thin Web Spiral Bevel Gear[J]. Acta Aeronautica et Astronautica Sinica, 2024, 45(12): 345-356.
[12] 刘敏, 杨扬, 左敦稳. 低频条件下二维振动应力释放规律研究[J]. 机械制造与自动化, 2021, 50(6): 56-59.
LIU M, YANG Y, ZUO D W.Research on Law of Residual Stress Relaxation of Two Dimensional Vibration at Low Frequency[J]. Machine Building & Automation, 2021, 50(6): 56-59.
[13] 刘昆, 喻立川, 向立果, 等. 基于最小二乘多项式拟合压力传感器补偿算法[J]. 自动化与仪器仪表, 2024(7): 52-56.
LIU K, YU L C, XIANG L G, et al.Pressure Sensor Compensation Algorithm Based on Least Square Method Polynomial Fitting[J]. Automation & Instrumentation, 2024(7): 52-56.
[14] 朱新颜, 孙鹏文, 孙文博, 等. 基于最小二乘多项式拟合的连续纤维基准铺放路径[J]. 复合材料科学与工程, 2025(5): 68-72.
ZHU X Y, SUN P W, SUN W B, et al.Continuous Fiber Reference Laying Path Based on Least Squares Polynomial Fitting[J]. Composites Science and Engineering, 2025(5): 68-72.
[15] 李永生, 陈子豪, 李丰润, 等. 非高斯随机振动疲劳寿命评估方法研究[J]. 机械, 2025, 52(3): 51-59.
LI Y S, CHEN Z H, LI F R, et al.Research on Non-Gaussian Random Vibration Fatigue Life Assessment Method[J]. Machinery, 2025, 52(3): 51-59.
[16] 李齐兵, 敬敏, 张梁娟, 等. 某机载单元随机振动疲劳分析[J]. 电子机械工程, 2018, 34(6): 32-35.
LI Q B, JING M, ZHANG L J, et al.Random Vibration Fatigue Life Analysis of an Airborne Unit[J]. Electro-Mechanical Engineering, 2018, 34(6): 32-35.
[17] 徐彬, 王腾, 林明辉, 等. 基于FEA与应力在线监测的RTG钢结构疲劳寿命预测[J]. 应用技术学报, 2024, 24(3): 316-320.
XU B, WANG T, LIN M H, et al.Fatigue Life Prediction of RTG Steel Structures Based on FEA and Stress Online Monitoring[J]. Journal of Technology, 2024, 24(3): 316-320.
[18] 陈卓, 闫明, 金映丽. 基于灰色马尔科夫模型的随机多次冲击损伤累积研究[J]. 振动与冲击, 2024, 43(24): 295-300.
CHEN Z, YAN M, JIN Y L.A Study on Damage Accumulation of Random Multiple Impact Based on a Grey Markov Model[J]. Journal of Vibration and Shock, 2024, 43(24): 295-300.
[19] BAKSHI A N, ALI BAIG MOGHAL A, MADHAR N A, et al. Effect of Stress on Energy Conversion and Storage Characteristics of (1-x-y)PIN-XPMN-yPT Single Crystals[J]. Ferroelectrics Letters Section, 2015, 42(4/5/6): 107-114.
[20] 代显智. 基于能量转换原理的磁电层合材料低频磁电响应分析[J]. 物理学报, 2014, 63(20): 314-323.
DAI X Z.Low Frequency Magnetoelectric Response Analysis of Magnetoelectric Laminate Material Based on Energy Conversion Principle[J]. Acta Physica Sinica, 2014, 63(20): 314-323.
[21] 杨燕, 贺启林, 朱海洋, 等. 带预载的钢丝网套补偿器振动疲劳损伤分析[J]. 机械工程学报, 2021, 57(15): 129-137.
YANG Y, HE Q L, ZHU H Y, et al.Vibration Fatigue Damage Analysis of Compensator Reinforced by Wire Cloth with Preload[J]. Journal of Mechanical Engineering, 2021, 57(15): 129-137.
[22] 谢嘉琳, 王志伟. 物流非高斯振动下的产品损伤分析[J]. 振动工程学报, 2024, 37(4): 645-656.
XIE J L, WANG Z W.Damage Analysis of Products under Logistics Non-Gaussian Vibration[J]. Journal of Vibration Engineering, 2024, 37(4): 645-656.
[23] 申加康, 徐俊, 张建军, 等. 装备振动功率谱密度统计归纳误差研究[J]. 振动与冲击, 2023, 42(17): 188-193.
SHEN J K, XU J, ZHANG J J, et al.Statistical Induction Error of Equipment Vibration Power Spectral Density[J]. Journal of Vibration and Shock, 2023, 42(17): 188-193.
[24] 腾振科技. 疲劳损伤计算: 基于 Q 值与 b 值的工程实践 [Z/OL]. (2022-12-08)[2025-10-17]. https://www.dtc-solutions.cn/solution/7426/7495.html.
[25] 腾振科技. 疲劳损伤测试: 基于频响函数的工程实践 [Z/OL]. (2023-04-11)[2025-10-17]. https://www.dtc-solutions.cn/solution/7426/7501.html.
Tengzhen Technology. Fatigue Damage Testing: Engineering Practice Based on Frequency Response Function [Z/OL]. (2023-04-11) [2025-10-17]. https://www.dtc-solutions.cn/solution/7426/7501.html.
PDF(1708 KB)
