滚动轴承声发射信号特征选取及状态识别方法研究

发布时间：2018-04-15 13:46

本文选题：声发射信号 + 滚动轴承　；参考：《北京化工大学》2012年硕士论文

【摘要】：本课题来源于国家自然科学基金项目《机械故障无线传感网络监测与智能诊断方法研究》（项目编号：51075023）。声发射检测技术用于故障诊断领域是当前研究的热点问题之一，声发射信号由于处于高频范围内，不易受到各种低频振动噪声干扰，具有良好的应用前景。但是，声发射检测技术在工业生产应用中，如何进行有效的信噪分离，如何选择特征参数以提高状态识别的准确率，如何选择适用于不同要求的状态识别方法是目前存在的三个主要问题。本文针对声发射监测系统中存在的主要问题开展研究，在信噪分离上提出了以小波变换为核心，以强制消噪法和小波阈值消噪法相结合的信号预处理方法，并将信息熵的概念引入小波基函数的选择过程中，选择出与滚动轴承声发射信号较为吻合的小波基函数，提高了消噪的效果；研究了基于现代信号处理方法的声发射信号特征提取方法，提出了以互信息理论和距离测度相结合的特征参数评价因子，选择敏感度高冗余度低的状态特征参数；研究了基于参数有效性的状态识别方法，提出特征参数与状态类别的相关性和参数之间距离测度作为有效性系数改进模糊识别方法，提高了识别的准确率。将改进的模糊聚类法与包络谱分析法作为监测单元简易状态识别方法，而将蚁群聚类算法和粒子群聚类算法作为上位机精确识别方法，，经过测试，这种监测单元简易识别与上位机精确识别结合的识别模式，提高了滚动轴承状态识别的效率。
[Abstract]:This topic comes from the project of National Natural Science Foundation "Research on Monitoring and Intelligent diagnosis of Mechanical Fault Wireless Sensor Network" (item No.: 51075023).Acoustic emission detection technology is one of the hot issues in the field of fault diagnosis. Acoustic emission signal is not easily disturbed by various low-frequency vibration noise because it is in the high frequency range, so it has a good application prospect.However, in the application of acoustic emission detection technology in industrial production, how to effectively separate signal and noise, how to select characteristic parameters to improve the accuracy of state recognition,How to choose a state recognition method suitable for different requirements is one of the three main problems.In this paper, the main problems in acoustic emission monitoring system are studied, and a signal preprocessing method based on wavelet transform is proposed, which combines the forced de-noising method with the wavelet threshold de-noising method.The concept of information entropy is introduced into the selection process of wavelet basis function and the wavelet basis function which is in good agreement with the acoustic emission signal of rolling bearing is selected to improve the effect of de-noising.The feature extraction method of acoustic emission signal based on modern signal processing method is studied. A feature parameter evaluation factor based on mutual information theory and distance measure is proposed to select the state characteristic parameters with high sensitivity and low redundancy.In this paper, the method of state recognition based on parameter validity is studied, and the correlation between feature parameter and state class and the distance measure between parameters are proposed as the effective coefficient to improve the fuzzy recognition method, which improves the accuracy of recognition.The improved fuzzy clustering method and envelope spectrum analysis method are used as simple state recognition methods for monitoring unit, while ant colony algorithm and particle cluster algorithm are used as the accurate identification method for upper computer.The combination of the simple identification of the monitoring unit and the precise identification of the upper computer improves the efficiency of the rolling bearing state recognition.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】