拉深件成形裂纹的非平稳信号处理及模糊识别研究

发布时间：2018-11-24 12:14

【摘要】：拉深裂纹作为金属板材成形件的主要失效形式,常在成形件危险区域出现,尤其是其中的早期裂纹,很难凭借传统的检测方法进行识别。针对上述所提到的问题,本研究采用声发射(AE)信号检测技术对金属成形件的拉深AE信号进行检测,再应用基于小波阀值-EMD综合法对裂纹AE信号进行分解降噪和重组,最后利用基于模糊等价关系的模糊聚类方法对各类裂纹进行模糊识别。主要的研究内容见下文:1)以盒形拉深件为本研究的理论模型,对金属板材成形件的拉深应力、应变状态和拉深过程出现裂纹的成因进行分析,然后结合金属板材拉深件畚斗的仿真结果得到此类拉深件容易出现拉深裂纹的危险区域;对无损检测方法中的声发射检测系统和工作原理进行介绍,在结合了金属板材拉深过程中裂纹的扩展伴随着声发射信号和声发射信号的特点,确定对金属板材拉深件进行深拉来获得拉深裂纹,并通过声发射检测系统对整个拉深过程进行监测以获得包含裂纹信号的声发射信号。2)结合上述理论分析,以金属板材拉深件畚斗为研究对象进行了金属板材拉深的AE信号采集实验,大量的采集和保存无裂纹、早起裂纹、扩展裂纹三种裂纹状态的AE信号,并对采集的信号和金属制件畚斗进行对应标注,以便于后期的数据处理。3)对声发射信号的降噪方法进行了研究,将其中对声发射降噪效果比较好的小波阈值滤波降噪、EMD降噪两种降噪方法进行分析和比较,并结合它们的优缺点提出了小波阈值-EMD综合降噪法,采用这种综合降噪法对金属板材拉深件的声发射信号进行降噪处理,在降噪前首先要采用消失矩为5的Daubechies小波基对AE信号进行三层小波分解操作,然后根据信号的频段进行频带选取,对选取的频带信号采用EMD降噪方法进行降噪,而没有选取的频带信号则采用小波阈值降噪,然后将两部分降噪过后的信号进行信号重构得到纯净的声发射信号。4)分析了金属板材成形裂纹产生的非平稳信号的特征,从降噪过后的纯净信号中提取出对应的信号参数;对模糊聚类算法进行介绍,选择基于模糊等价关系的模糊聚类方法对四类信号进行模糊识别,选取相互独立的幅度、有效值电压(RMS)、平均信号电平值(ASL)、能量四个参数作为识别参数,建立数据矩阵,运用MATLAB软件对金属板材拉深件畚斗的提取参数进行数值模拟分析,实现对无裂纹、早起裂纹和扩展裂纹三种状态裂纹的多参数模糊识别。研究结果表明:利用声发射采集系统采集金属制件畚斗的声发射信号,采用小波阈值-EMD综合降噪法对拉深信号进行降噪处理和重构,然后从中提取特征信号作为模糊聚类的参数,实现了对金属板材拉深件畚斗裂纹状态(尤其是早期裂纹)的模糊识别,且识别准确率较高。
[Abstract]:Drawing cracks, as the main failure forms of sheet metal forming parts, often appear in the dangerous areas of forming parts, especially the early cracks, so it is difficult to identify them by traditional detection methods. In order to solve the problems mentioned above, the acoustic emission (AE) signal detection technique is used to detect the deep drawing AE signal of metal forming parts, and then the cracked AE signal is decomposed and recombined based on wavelet threshold EMD synthesis method. Finally, the fuzzy clustering method based on fuzzy equivalence relation is used to identify all kinds of cracks. The main research contents are as follows: 1) taking the box drawing parts as the theoretical model of this study, the paper analyzes the drawing stress, strain state and the causes of cracks in the drawing process of sheet metal forming parts. Then combining with the simulation results of the metal sheet drawing bucket, the dangerous area where the drawing crack is easy to appear is obtained. This paper introduces the acoustic emission testing system and working principle in the nondestructive testing method. It combines the characteristics of acoustic emission signal and acoustic emission signal in the process of metal sheet drawing. It is determined that deep drawing is carried out to obtain deep drawing crack of metal sheet drawing, and the whole drawing process is monitored by acoustic emission detection system to obtain acoustic emission signal containing crack signal. 2) combined with the above theoretical analysis, Taking the metal plate deep drawing bucket as the research object, the AE signal acquisition experiment of metal plate drawing is carried out. A large number of AE signals are collected and preserved in three kinds of crack states, that is, no crack, early rise crack and propagating crack. At the same time, the signal and metal dustbin are labeled accordingly, so as to facilitate the later data processing. 3) the noise reduction method of acoustic emission signal is studied, and the wavelet threshold filter, which has better effect on acoustic emission noise reduction, is used to reduce the noise. Two methods of EMD noise reduction are analyzed and compared. Combined with their advantages and disadvantages, the wavelet threshold-EMD comprehensive noise reduction method is put forward, which is used to reduce the noise of the acoustic emission signal of the metal sheet drawing parts. Before noise reduction, Daubechies wavelet basis with vanishing moment of 5 is used to decompose AE signal with three-layer wavelet transform, then the frequency band is selected according to the frequency band of the signal, and the selected frequency band signal is de-noised by EMD denoising method. The non-selected band signal is de-noised by wavelet threshold, and then the two parts of the de-noised signal are reconstructed to get the pure acoustic emission signal. 4) the characteristics of the non-stationary signal produced by the metal sheet forming crack are analyzed. The corresponding signal parameters are extracted from the pure signal after noise reduction. This paper introduces the fuzzy clustering algorithm, selects the fuzzy clustering method based on the fuzzy equivalence relation to carry on the fuzzy recognition to the four kinds of signals, selects the mutually independent amplitude, the effective value voltage (RMS), average signal level value (ASL), Four parameters of energy are used as identification parameters, data matrix is established, and the extraction parameters of metal plate deep drawing bucket are numerically simulated and analyzed by using MATLAB software, and the crack free is realized. Multi-parameter fuzzy identification of early crack and propagating crack. The results show that the acoustic emission signal of metal dustbin is collected by acoustic emission acquisition system, and the deep drawing signal is de-noised and reconstructed by wavelet threshold-EMD comprehensive de-noising method. Then the feature signal is extracted as the parameter of fuzzy clustering to realize fuzzy identification of bucket crack state (especially early crack) of metal sheet drawing parts and the accuracy of identification is high.
【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG386.32

【参考文献】