【作者】 李艳兰；

【导师】 熊诗波； 熊晓燕；

【作者基本信息】 太原理工大学 ， 机械电子工程， 2005， 硕士

【摘要】 随着电能应用的方便,电机设备已被广泛应用于工业生产的各个领域。电机在整个机械系统中起着举足轻重的作用,一旦电机发生故障就会影响整个系统的正常运行,甚至危及人身安全。所以对电机进行故障诊断非常重要。 本文在对异步电动机常见电气故障机理分析的基础上,提出了利用定子电流法来诊断电机电气故障。并对发生率比较高的气隙偏心和匝间短路故障进行了重点研究。 因为电机内部存在电路和磁路的耦合,所以电机故障诊断涉及的因素很多。相应地,电机故障诊断技术比一般的机械设备诊断技术涉及范围更广。电机故障诊断的复杂性还表现在故障特征量的隐含性、故障起因与故障征兆之间的多元性。即一个故障可能表现为多种故障征兆;有时不同故障起因也可能表现为同一个故障征兆。针对电机故障的特点,本文提出小波与神经网络相结合的方法,利用“小波包—能量”法提取出信号特征量,作为神经网络的输入向量,采用包括更多还原

【Abstract】 For the convenience that is widely use with the electric energy, the electric motor equipment has already widely used in each industrial field. The electric motor plays an important role in the whole mechanical systems. It will influence the normal running of the whole system if the electric motor has fault, even threaten the security of the person. So it is very important to carry on trouble diagnosis for electric motor.On the basis of analysing the common electric trouble mechanism of induction motor, the paper proposes making use of stator current law to diagnose the induction motor’s electric fault. Furthermore it gives the priority research to the airgap eccentricity and turn-to-turn short circuit fault with high incidence rate.Because the coupled electric circuit and magnetic circuit exist in the induction motor, there are many factors of the motor fault. Accordingly, the range of the induction motor trouble diagnosis technology is more widely than the general mechanical equipment technology. The complexity of induction motor also represents the implicity of the fault characteristics, plurality between the cause and sign of the fault. That is to say, a trouble may present many kinds of trouble signs while different troubles may present the same trouble sign. Aimed to the characteristics of the trouble of the induction motor, this paper puts forward the method that the wavelet combines with neural network, and picks up the character vector by using the wavelet packet and energy method as the ANN input neural vector. The ANN is trained by teacher using LM algorithm. Not having to establish the model structure of the system, this method makes use of energy change of the frequency composition to discern different modes.Because it is inaccurate to measure the stator current of induction motor by the current transformer, the stator current data gets through measuring the voltage of wire from Power to induction motor. The current of the wire and the stator current are the same electric current, so the stator current is gotten by measuring the wire current. The experiment testifies that it is feasibility that applying the wavelet packet and neural network to the electric fault diagnosis of the induction motor.At last, the paper uses the strong function of Graphical User Interfaces of MATLAB to combine the signal processing and neural network technology. It is convenience to the users analyzing the signal.更多还原