基于BO-WT-VMD组合模型的钢丝绳的损伤信号降噪方法
Noise Reduction Method for Damage Signal of Wire Rope Based on BO-WT-VMD Combined Model
朱宇坤 1黎恒 2吴文俊 2柯圆圆2
作者信息
- 1. 江汉大学 智能制造学院,湖北 武汉 430056
- 2. 江汉大学 人工智能学院,湖北 武汉 430056
- 折叠
摘要
为了准确识别钢丝绳缺陷的故障数据,消除被测信号中的噪音,可使用基于贝叶斯优化-小波变换-变分模态分解(BO-WT-VMD)的组合降噪模型进行降噪.首先利用小波变换(WT)对原始信号进行处理;其次对处理后的信号再利用变分模态分解算法(VMD)进行模态分解,计算分解的各个模态与原信号之间的相关系数,以完成信号的模态选择和重构;再次利用贝叶斯优化算法(BO)对VMD算法的模态数目K以及α值进行寻参,避免其受序列分解和预测精度的影响,从而找到最优超参数组合.最后,在仿真实验和真实数据中,计算BO-WT-VMD、WT和VMD算法的SNR值和RMSE值,分别为(17.340 3,0.101 0),(15.617 0,0.123 2)和(15.649 2,0.122 7),说明提出的算法能够有效地去除信号噪音.
Abstract
To accurately identify the fault data of wire rope defects and eliminate the noise in the measured signal,this paper used a combined noise reduction model based on the Bayesian optimization-wavelet transform-variational moda decomposition(BO-WT-VMD)for noise reduction.Firstly,the wavelet transform(WT)was used to process the original signal.Then,the variational mode decomposition(VMD)was used for the modal decomposition of the signal processed by the wavelet transform.The correlation coefficient between the decomposed mode and the original signal was calculated to complete the signal mode selection and reconstruction.Finally,the Bayesian optimization(BO)algorithm was used to search for parameters K and the value α in the VMD algorithm,avoiding the impact of sequence decomposition and prediction accuracy,in order to find the optimal hyperparameter combination.Furthermore,in simulation trials and real data,the SNR and RMSE of the BO-WT-VMD,WT,and VDM algorithms are calculated respectively,which were(17.340 3,0.101 0),(15.617 0,0.123 2),(15.649 2,0.122 7),indicating that the algorithm proposed in this paper can effectively eliminate signal noise.
关键词
磁漏检测/小波变换/变分模态分解算法/贝叶斯优化/钢丝绳/降噪Key words
magnetic leakage detection/wavelet transform/variational mode decomposition/Bayesian optimization/wire rope/noise reduction引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据