基于EMD和FastICA算法的齿轮箱故障诊断研究

发布时间：2018-04-19 12:16

本文选题：齿轮箱 + EMD　；参考：《华北水利水电大学》2017年硕士论文

【摘要】：齿轮箱具有结构紧凑,承载能力强,传动效率高,可靠性强等优点,能够满足高速大功率及低速大扭矩的传动要求,是各种机械动力传动系统的重要装置,其工作状态是否正常对运动和动力的传输具有重要影响,一旦发生事故,将会造成严重影响,因此基于EMD和FastICA算法对齿轮箱故障诊断进行研究具有重要的理论价值和现实意义。本文的主要研究内容是基于EMD和FastICA算法的单通道盲源分离方法在齿轮箱故障诊断的应用。在理论方面,主要分析了齿轮的振动机理和滚动轴承的振动机理,研究了齿轮箱中齿轮、轴承、轴和箱体的若干个典型故障,包括断齿、齿面磨损、齿轮误差、齿轮偏心、外圈故障、内圈故障、保持架故障、滚动体故障、轴不对中、轴不平衡、箱体共振,建立了对齿轮箱故障的整体认识,为后期进行齿轮箱的故障诊断奠定理论基础。在实验方面,在齿轮箱动力模拟系统设备的基础上,人为设置齿轮箱各种不同工况(齿轮箱正常、齿轮齿面磨损故障、缺齿故障和复合故障),通过压电式加速度传感器和激光转速传感器将齿轮箱的振动信号传递到计算机上的HG8916综合数据采集故障诊断系统,利用HG8916的数据采集模块中的时域数据采集模块采集信号,并观察信号的时域特征,然后将HG8916采集出的信号导出,转换为txt格式。针对单一通道信号,在MATLAB中编写程序,调用EMD程序将信号进行分解成IMF分量,根据改进的奇异值分解法进行源数估计,重组虚拟信号,再调用FastICA算法对信号进行处理,提取盲源分离矩阵的奇异值作为信号故障特征,通过BP神经网络进行分类识别和判断,得出故障诊断的结果。实验取得了良好的效果。通过实验验证可以得出以下结论:基于EMD和Fast ICA算法的单通道盲源分离方法综合了EMD算法和FastICA算法的优点,原理和计算相对简单,且操作方便,不存在人为主观因素,非常适合齿轮箱故障诊断的应用。
[Abstract]:The gearbox has the advantages of compact structure, strong bearing capacity, high transmission efficiency and high reliability. It can meet the requirements of high speed, high power and low speed and large torque, and is an important device of various mechanical power transmission systems.Whether its working state is normal or not has an important influence on the transmission of motion and power. Once an accident occurs, it will cause serious impact. Therefore, the research of gearbox fault diagnosis based on EMD and FastICA algorithm has important theoretical value and practical significance.The main research content of this paper is the application of single channel blind source separation method based on EMD and FastICA algorithm in gearbox fault diagnosis.In theory, the vibration mechanism of gear and the vibration mechanism of rolling bearing are analyzed, and some typical faults of gear, bearing, shaft and box in gear box are studied, including tooth breaking, tooth surface wear, gear error, gear eccentricity.Outer ring fault, inner ring fault, cage fault, rolling body fault, shaft misalignment, shaft imbalance, box resonance, established the overall understanding of the gearbox fault, and laid a theoretical foundation for the fault diagnosis of the gearbox in the later stage.In the aspect of experiment, on the basis of the gear box dynamic simulation system equipment, the gear box is artificially set up various different working conditions (gear box normal, gear tooth surface wear failure,The vibration signal of gearbox is transmitted to the HG8916 integrated data acquisition fault diagnosis system by piezoelectric accelerometer and laser speed sensor.The time-domain data acquisition module of HG8916 is used to collect signals and observe the time-domain characteristics of the signals. Then the signals collected by HG8916 are exported and converted into txt format.For single channel signal, program is written in MATLAB, EMD program is called to decompose the signal into IMF component, the source number is estimated according to the improved singular value decomposition method, the virtual signal is reorganized, and then the FastICA algorithm is called to process the signal.The singular value of the blind source separation matrix is extracted as the signal fault feature, and the fault diagnosis results are obtained by BP neural network.The experiment has achieved good results.The experimental results show that the single channel blind source separation method based on EMD and Fast ICA algorithm combines the advantages of EMD algorithm and FastICA algorithm. The principle and calculation are relatively simple, the operation is convenient, and there are no artificial subjective factors.It is very suitable for gearbox fault diagnosis.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH132.41

【参考文献】