磁声发射信号样本熵特征量的应力检测方法研究
Research on stress detection method based on sample entropy characteristic of magnetic acoustic emission signal
陈炫宇 1龙盛蓉 1韩壮禄 1陈学宽 1李志农1
作者信息
- 1. 南昌航空大学江西省图像处理与模式识别重点实验室,江西南昌 330063;南昌航空大学无损检测教育部重点实验室,江西南昌 330063
- 折叠
摘要
为解决在应力检测中传统磁声发射(magnetic acoustic emission,MAE)特征量易受噪声干扰的问题,提出基于磁声发射信号样本熵特征量的应力检测方法.在研究Q235钢试样在 0～400 MPa拉伸应力状态下磁声发射信号中样本熵变化规律的基础上,分析励磁强度、嵌入维数、容限系数对磁声发射信号样本熵特征量的影响规律.结果表明磁声发射信号样本熵特征量与应力具有良好的对应关系,并且相较于传统时域特征量(峰峰值和均方根值),样本熵值的波动性指标分别下降 89%和 33%,在所选励磁条件下由样本熵值计算得到的应力值和实际应力值平均误差仅为8.5%,其受噪声干扰较小且一致性更好,更适合应用于铁磁性材料的应力检测.
Abstract
In order to solve the problem that the characteristic quantity of traditional magnetic acoustic emission(MAE)signal is easily disturbed by noise in stress detection,a stress detection method based on sample entropy eigenvalues of magnetic acoustic emission signals is proposed.Based on the study of the variation law of sample entropy in magnetoacoustic emission signal of Q235 steel sample under 0-400 MPa tensile stress,the effects of excitation intensity,embedding dimension and tolerance coefficient on the characteristic quantity of sample entropy of magnetic acoustic emission signal are analyzed.The results show that the sample entropy characteristic quantity of magnetic acoustic emission signal has a good corresponding relationship with stress,and compared with the traditional time-domain characteristic quantity(peak to peak value and root mean square value),the fluctuation index of sample entropy decreases by 89%and 33%respectively,the average error between the stress value calculated from the sample entropy value and the actual stress value is only 8.5%under the selected excitation conditions.It is less disturbed by noise and has better consistency,which is more suitable for the stress feature extraction of magnetic acoustic emission signal.
关键词
样本熵/应力特征/铁磁性材料/磁声发射Key words
sample entropy/stress characteristics/ferromagnetic materials/magnetic acoustic emission引用本文复制引用
基金项目
国家自然科学基金(52075236)
装备预研基金(6142003190210)
江西省图像处理与模式识别重点实验室开放基金(ET202008414)
出版年
2024
NETL
NSTL
万方数据