摘要:对电机常见故障的在线诊断和分析，不仅可以及早地发现故障和预防故障的进一步恶化，减少突发事故造成的停产损失，防止对人员和设备安全的威胁，并为实现状态检修创造条件；而且能为设计制造者提供经验，积累数据，有利于电机性能及可靠性的改进；同时对于电机故障的定位、类型决策及其维修等都是极其重要的。

　　　论文在对电机常见故障类型及故障发生的机理进行详细分析的基础上，提出了基于BP和RBF神经网络的电机故障诊断方案。主要内容包括：

　　　论文首先详细分析了电机常见故障的类型，研究了包括轴承故障、转子偏心故障、电刷故障、电枢故障等几种常见故障的产生机理及特征。

　　　然后在分析了BP神经网络和RBF神经网络的基本原理的基础上，讨论了神经网络应用于电机故障诊断的可行性。由于电机发生故障时，定子电流的幅值在其相对应的特征频率上的会出现十分明显的增加，而且这些频率点上的幅值所增加的大小是与故障的严重程度成正比。利用这一信号特征，我们就可以对故障进行判断和分类。

　　　最后，利用上述信号特征，论文分别提出了基于BP神经网络算法和基于RBF神经网络算法的电机故障诊断方案，并通过实验验证了该算法的正确性。

关键词：神经网络；电机故障；诊断

Abstract:Common motor online fault diagnosis and analysis, not only for early failure and the further deterioration of the prevention of failure, reducing the cut-off losses caused by unexpected incidents, to prevent a threat to the  safety of personnel and equipment, and create conditions for realization of state maintenance; but also to provide experience for the design of the manufacturer, the accumulation of data, is conducive to the improvement of motor performance and reliability; for the positioning of the motor failure, types of decision-making and its maintenance are extremely important.

　　On the basis of the analysis of the mechanism of motor common type fault and failure, the motor fault diagnosis based on BP and RBF neural network is researched. The main contents include:

　　Firstly, the common type failure of motor is analyzed, and the generation mechanism and characteristics of several common failures, such as rotor eccentricity fault, brush failure, armature failure, are researched.

　　And then on the basis of analysis of the basic principle of BP neural network and RBF neural network to discuss the feasibility of neural network used in motor fault diagnosis. Due to motor failure, the stator current amplitude will appear in its characteristic frequency corresponding to the apparent increase in the size of the increase in the amplitude of these frequency points is proportional to the severity o f the fault. The characteristics of this signal, we can judge and classify the fault.

　　Finally, using the above signal characteristics, the methods based on BP neural network algorithm and RBF neural network algorithm for motor fault diagnosis are introduced, and the correctness of the algorithm is verified by experiments.

Key words: Neural network；Motor fault；Diagnosis