基于小波包和最小二乘支持向量机的异步电机故障诊断

发布时间：2018-05-03 20:10

本文选题：鼠笼异步电机 + 故障诊断　；参考：《中国矿业大学》2014年硕士论文

【摘要】：鼠笼异步电机因为其简单的结构、可靠的运行以及低廉的价格，在工农业生产中得到广泛应用。但是受恶劣的工作环境、频繁的起动制动以及外力等其他一些因素的影响，异步电机常常会发生各种各样的故障。这些电机故障不仅会损坏电机本身，而且还会影响整个工业生产的运行。因此，能够使用行之有效的方法对常见故障进行诊断和分析是当前研究鼠笼型异步电机工作的主流。传统的诊断方法存在一定局限性，不能很好地进行故障诊断。文章以鼠笼异步电机的定子电流信号为研究对象，以定子电流中的特征频率为视角，提出了以小波包理论和支持向量机理论为基础的故障诊断方法。主要工作如下：首先在分析鼠笼电机转子基本结构，，和转子断条故障原因、故障机理基础上，对诊断方法进行了总结，得到了定子电流诊断方法；其次学习了小波变换和小波包变换的基本原理，并进一步提出了基于小波包理论的最优小波包基降噪和频带能量方法，并使用MATLAB对该方法的有效性和可靠性进行了仿真验证。再次，概要分析了机器学习的基本理论，着重对分类支持向量机理论以及其改进理论最小二乘支持向量机进行了分析与理论推导；最后以从故障诊断试验平台采集到的定子电流信号为原始数据，采用上面的方法通过编写MATLAB程序逐步分析，得到了较高的故障诊断准确率。文章认为以小波包和最小二乘支持向量机为基础的方法能够很好的消除噪声、提取特征信息、进行准确的分类判断。具有很好理论推广价值和应用价值。
[Abstract]:Squirrel cage asynchronous motor is widely used in industry and agriculture because of its simple structure, reliable operation and low price. However, due to the bad working environment, frequent starting and braking, and other factors such as external force, asynchronous motors often have a variety of faults. These motor failures will not only damage the motor itself, but also affect the operation of the entire industrial production. Therefore, it is the mainstream of the research on squirrel cage asynchronous motor to be able to diagnose and analyze the common faults by using effective methods. The traditional diagnosis method has some limitations and can not be used well in fault diagnosis. In this paper, a fault diagnosis method based on wavelet packet theory and support vector machine (SVM) is proposed based on the stator current signal of squirrel cage asynchronous motor and the characteristic frequency of stator current. The main tasks are as follows: Firstly, on the basis of analyzing the basic structure of squirrel cage motor rotor, the cause of rotor broken bar fault and the fault mechanism, the diagnosis method is summarized, and the stator current diagnosis method is obtained. Secondly, the basic principles of wavelet transform and wavelet packet transform are studied, and the optimal wavelet packet basis denoising and frequency band energy method based on wavelet packet theory is proposed, and the effectiveness and reliability of the method are verified by using MATLAB. Thirdly, the basic theory of machine learning is analyzed, and the classification support vector machine theory and its improved least square support vector machine theory are analyzed and deduced. Finally, taking the stator current signal collected from the fault diagnosis test platform as the original data, the method above is used to analyze the stator current signal by writing the MATLAB program step by step, and the higher accuracy of fault diagnosis is obtained. This paper holds that the method based on wavelet packet and least square support vector machine can eliminate noise, extract feature information and make accurate classification and judgment. It has very good theory popularization value and application value.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM343

【参考文献】