多传感器信息融合技术在电机故障诊断中的应用研究

发布时间：2018-07-10 14:39

本文选题：电机 + 故障诊断　；参考：《兰州理工大学》2014年硕士论文

【摘要】：电机是最主要的机电能量转换设备,不论是在国民经济中的各种能源、制造领域里,还是在人们的日常生活中,电机都有着无可替代的地位。电机故障诊断技术的研究具有重大的经济意义和社会意义,经历了几十年的发展,电机故障诊断技术取得了长足进步,不论是在信号处理方面还是在诊断方法上与发展之初相比都己不可同日而语,然而,目前普遍采用的基于单参数、单特征的电机故障诊断系统在诊断过程中仍存在很大的不确定性,有时往往难以保证诊断的精确度,在此基础上,本文构建了一种基于多传感器信息融合技术的电机故障诊断方法。本文以电机故障诊断为研究对象,首先介绍了电机故障诊断技术的背景、意义及发展,分析了电机故障诊断技术的发展趋势,同时对多传感器信息融合技术进行了简介,并对电机定子故障、转子故障、轴承故障及气隙偏心故障等常见故障进行了分析。在信号处理及特征提取方面,针对希尔伯特-黄变换的核心内容经验模态分解中存在的模态混叠以及虚假分量问题进行了改进。通过仿真实验,验证了集合经验模态分解在抑制模态混叠现象时的可行性；采用了利用灰色关联度进行虚假分量识别的方法,通过与相关系数法的对比仿真,证明了灰色关联度在识别虚假分量时的有效性。并在此基础上,利用各固有模态函数能量构造故障特征向量。在故障局部诊断方法上,采用了目前应用最为广泛、理论最为成熟的BP神经网络,介绍了神经网络的基础知识、原理、结构及学习过程,利用神经网络的优良特性为后续的基于D-S证据理论的信息融合方法提供精度和可靠性更高的输入信息。在信息融合算法方面,对D-S证据理论的基本概念及D-S合成规则进行了介绍和分析,在基本可信度分配函数的构成上,利用神经网络局部诊断结果作为基础综合考虑误差因素,不仅解决了D-S证据理论中如何构建基本可信度分配函数的难点,同时也避免了D-S合成规则难以处理冲突证据的缺陷,将神经网络与D-S证据理论有效结合起来。最后,本文构建了基于多传感器信息融合技术的电机故障诊断系统模型,并选择电机故障中最为常见的轴承故障作为实验对象,对诊断系统进行了实验验证和数据分析。通过试验,验证了本文所构建的基于多传感器信息融合的电机故障诊断系统具有可行性、正确性和有效性。
[Abstract]:Motor is the most important electromechanical energy conversion equipment, whether in the national economy in all kinds of energy, manufacturing field, or in the daily life of people, the motor has an irreplaceable position. The research of motor fault diagnosis technology has great economic and social significance. After decades of development, the motor fault diagnosis technology has made great progress. Neither in signal processing nor in diagnostic methods have been compared with the beginning of its development. However, it is widely used at present on the basis of single parameter. The single feature motor fault diagnosis system still has a lot of uncertainty in the process of diagnosis, and sometimes it is difficult to guarantee the accuracy of the diagnosis. In this paper, a method of motor fault diagnosis based on multi-sensor information fusion technology is proposed. In this paper, motor fault diagnosis is taken as the research object. Firstly, the background, significance and development of motor fault diagnosis technology are introduced, and the development trend of motor fault diagnosis technology is analyzed. At the same time, the multi-sensor information fusion technology is introduced briefly. The common faults such as stator fault, rotor fault, bearing fault and air gap eccentricity fault are analyzed. In the aspect of signal processing and feature extraction, the problems of modal aliasing and false components in empirical mode decomposition of Hilbert-Huang transform are improved. Through the simulation experiment, the feasibility of the set empirical mode decomposition in suppressing the mode aliasing phenomenon is verified, and the method of using grey correlation degree to identify the false component is adopted, and the simulation results are compared with the correlation coefficient method. The validity of grey correlation degree in identifying false components is proved. On this basis, the fault eigenvector is constructed by using the energy of each inherent mode function. In the method of fault local diagnosis, BP neural network, which is the most widely used and the most mature theory at present, is adopted. The basic knowledge, principle, structure and learning process of neural network are introduced. The advantages of neural network can provide more accurate and reliable input information for the subsequent information fusion method based on D-S evidence theory. In the aspect of information fusion algorithm, the basic concept of D-S evidence theory and D-S synthesis rule are introduced and analyzed. In the structure of basic reliability distribution function, the error factors are synthetically considered based on the local diagnosis result of neural network. It not only solves the difficulty of how to construct the basic reliability assignment function in D-S evidence theory, but also avoids the defect that D-S synthesis rule is difficult to deal with conflict evidence, and combines neural network and D-S evidence theory effectively. Finally, a fault diagnosis system model of motor based on multi-sensor information fusion technology is constructed, and the most common bearing fault in motor fault is selected as the experimental object, and the experimental verification and data analysis of the diagnosis system are carried out. The experimental results show that the motor fault diagnosis system based on multi-sensor information fusion is feasible, correct and effective.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM307

【参考文献】