基于改进提升小波的AE信号消噪及在故障诊断中的应用研究

发布时间：2018-01-12 23:33

本文关键词：基于改进提升小波的AE信号消噪及在故障诊断中的应用研究　出处：《湖南科技大学》2012年硕士论文　论文类型：学位论文

【摘要】：旋转机械是国民生产中的关键核心设备，对其进行监测与诊断，保障其安全、可靠运行，对提高企业的经济效益、促进整个国民经济的发展有着重要的作用。传统的监测与诊断一般基于振动信号，而振动信号分析法在强噪声干扰背景下难以准确获取设备的故障特征信息。基于声发射(AE)信号的监测与诊断方法具有灵敏度高、抗噪声干扰能力强等优点，更易于提取可表征设备运行状态的特征信息，将其应用于旋转机械故障诊断，可以有效的提高故障诊断的准确率。本文在分析以滚动轴承为代表的旋转机械声发射信号特性的基础上，针对传统提升小波的不足，提出了一种适合于旋转机械AE信号的改进提升小波变换方法。研究了基于改进提升小波的旋转机械AE信号消噪方法，并结合经验模态分解开展AE信号特征提取、结合BP神经网络开展AE信号故障识别研究，具体研究工作如下： (1)提出了旋转机械AE信号提升小波改进算法。通过理论分析，结合实验研究探明旋转机械声发射信号特性，在此基础上提出改进提升小波，包括分解过程中，引入局部判决函数，根据被分析AE信号的局部特征自适应的构造更新算子；利用遗传算法，，对预测算子进行优化。重构过程中，利用自适应阈值消噪对每一层的AE信号进行消噪，提高了提升小波的自适应消噪性能。 (2)研究了基于改进提升小波的旋转机械AE信号消噪方法。针对自适应阈值消噪公式中参数p的取值过大或过小都会影响消噪的效果，通过AE仿真信号研究了改进提升小波在AE信号消噪中的最佳参数值，并用实测的AE信号验证了改进提升小波比传统提升小波和小波变换具有更好的消噪性能。 (3)采用改进提升小波和经验模态分解对旋转机械故障的AE信号进行了特征提取研究。利用改进提升小波对经验模态分解的AE信号进行消噪处理，然后用相关函数法求出经验模态分解后的有效内禀模态函数，并进行包络解调分析，能够准确提取旋转机械AE信号中的故障特征信息。 (4)结合改进提升小波和BP神经网络对滚动轴承进行了故障识别研究。将改进提升小波作为BP神经网络的前处理器，对AE信号进行消噪预处理，提取消噪后AE信号的特征参数作为BP神经网络的输入量，减少了BP神经网络的训练步骤，提高了轴承故障识别的准确率。
[Abstract]:The rotating machinery is the key core equipment in the national production . It plays an important role in monitoring and diagnosing it , ensuring the safety and reliable operation of the enterprise . The traditional monitoring and diagnosis is based on the vibration signal , and the vibration signal analysis method has the advantages of high sensitivity and strong anti - noise interference . It is easier to extract the characteristic information which can characterize the running state of the equipment . It can be applied to the fault diagnosis of rotating machinery , which can effectively improve the accuracy of fault diagnosis . On the basis of analyzing the characteristics of the rotating machinery acoustic emission signal represented by the rolling bearing , this paper presents an improved wavelet transform method which is suitable for the rotating mechanical AE signal . Based on the improvement of the wavelet transform method , the AE signal feature extraction is carried out in combination with the empirical mode decomposition , and the AE signal feature extraction is carried out in combination with the BP neural network . The specific research work is as follows : ( 1 ) An improved algorithm for improving the AE signal of a rotating machinery is proposed . Based on the theoretical analysis , the characteristics of the acoustic emission signal of the rotating machinery are investigated by combining the experimental study . On the basis of this , an improved wavelet is proposed , which includes the introduction of a local decision function in the decomposition process , the optimization of the predictive operator based on the local feature of the analyzed AE signal . In the reconstruction process , the adaptive threshold de - noising is used to de - noising each layer of AE signal , which improves the adaptive noise elimination performance of the lifting wavelet . ( 2 ) The method of improving the signal denoising based on the modified lifting wavelet is studied . The optimal parameter value of the wavelet in the noise elimination of AE signal is studied by AE simulation signal . The best parameter value of the wavelet in the noise elimination of AE signal is studied through AE simulation signal . The improved wavelet and wavelet transform are improved with the measured AE signal to improve the noise elimination performance . ( 3 ) An improved wavelet and empirical mode decomposition is used to extract the AE signal of the rotating machinery fault . An improved wavelet is used to denoise the AE signal decomposed by the empirical mode , then the effective intrinsic mode function after the empirical mode decomposition is obtained by the correlation function method , and the envelope demodulation analysis is carried out to accurately extract the fault feature information in the rotating mechanical AE signal . ( 4 ) Combining the improved lifting wavelet and BP neural network to study the fault identification of the rolling bearing , the improved lifting wavelet is used as the front processor of the BP neural network , the noise elimination preprocessing is carried out on the AE signal , the characteristic parameter of the AE signal after canceling noise is taken as the input amount of the BP neural network , the training step of the BP neural network is reduced , and the accuracy of the bearing fault identification is improved .

【学位授予单位】：湖南科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TN911.4;TH165.3

【参考文献】