滚动轴承故障机理及智能化检测技术研究

发布时间：2018-12-17 19:13

【摘要】：制造业是科技革新的动脉,高端设备制造业更是关乎一个国家社会安全、国防安全和经济安全的新兴战略产业。随着以滚动轴承为代表的旋转机械零部件的广泛应用,制造业发展革新也面临着设备维护、机械检修、寿命预测、状态识别等一系列问题。针对不同原因、不同表征、不同程度的机械故障的智能化检测技术已成为当前的研究热点和难点。本课题以滚动轴承为研究对象,基于故障机理知识,开展信号采集、信号分析、信号处理等技术的研究,实现智能化检测,具体内容如下:(1)阐述滚动轴承的结构特性、分类形式和失效种类。从故障机理的角度,对滚动轴承进行运动学分析,分别对滚动轴承动力学、故障荷载、故障信号进行建模。与此同时,研究讨论故障出现的位置对振动信号特征的影响。(2)基于信号处理和机器学习方法的先验知识,分别提出基于自回归和谱熵算法的检测技术和基于字典和增强学习算法的检测技术。前者采用自回归模型剔除机械振动信号中可线性预测的平稳成分,在不同频带下进行复Morlet小波包络,结合谱熵在频域内与通带滤波的相关性选定最优包络。后者采用字典学习和稀疏编码对机械振动信号进行预处理,并引入反向传播神经网络构建自适应增强学习的分类函数,采用机器学习方法训练集实现智能化检测。基于仿真和试验结果验证提出技术的有效性和先进性,在工程应用中具有良好的前景。(3)从试验台机械结构设计、控制系统运行通信以及软件开发三个方面搭建检测平台。参与电机和滚动轴承的选型,加速度传感器的安装及信号采集,控制系统的编程,数据可视化,数值计算分析,人机交互界面设计等工作。将故障机理和提出技术应用于滚动轴承故障检测平台,试验效果显著。
[Abstract]:Manufacturing industry is the artery of scientific and technological innovation. High-end equipment manufacturing industry is related to a country's social security, national defense security and economic security emerging strategic industries. With the extensive application of rotating machinery parts represented by rolling bearings, the development and innovation of manufacturing industry is faced with a series of problems, such as equipment maintenance, mechanical maintenance, life prediction, state recognition and so on. According to different reasons, different representations, different degrees of intelligent detection technology of mechanical faults has become a hot and difficult research. This subject takes rolling bearing as the research object, based on the knowledge of fault mechanism, carries out the research of signal acquisition, signal analysis, signal processing and so on, and realizes intelligent detection. The specific contents are as follows: (1) expound the structural characteristics of rolling bearing. Classification form and failure type. From the point of view of fault mechanism, the kinematics analysis of rolling bearing is carried out, and the dynamics, fault load and fault signal of rolling bearing are modeled respectively. At the same time, the influence of fault location on vibration signal characteristics is discussed. (2) A priori knowledge based on signal processing and machine learning, The detection techniques based on autoregressive and spectral entropy algorithms and dictionary and enhanced learning algorithms are proposed respectively. The former uses autoregressive model to eliminate the linear predictable stationary components of mechanical vibration signals and carries out complex Morlet wavelet envelopes in different frequency bands. The optimal envelope is selected in combination with the correlation between spectral entropy and passband filtering in frequency domain. The latter uses dictionary learning and sparse coding to preprocess mechanical vibration signal, and introduces back propagation neural network to construct classification function of adaptive reinforcement learning. Machine learning method training set is used to realize intelligent detection. Based on the simulation and experimental results, the effectiveness and advanced nature of the proposed technology are verified, and it has a good prospect in engineering application. (3) the testing platform is built from three aspects: the mechanical structure design of the test-bed, the running communication of the control system and the software development. It is involved in the selection of motor and rolling bearing, the installation of acceleration sensor and signal collection, the programming of control system, data visualization, numerical calculation and analysis, and the design of man-machine interface. The fault mechanism and the proposed technology are applied to the rolling bearing fault detection platform, and the test results are remarkable.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH133.33

【参考文献】